Data Science Projects for Freelancers

We are looking for a talented team of professionals to build, label, and deploy Decathlon's AI image system. The project requires a Machine Learning Engineer to develop algorithms, a Data Annotation Specialist to accurately label data, and a Web Developer to create an intuitive interface. Collaborate with us to enhance our AI capabilities and provide a seamless user experience. If you have experience in these areas and are passionate about AI and technology, we would love to hear from you!

I’m building a software tool that can break down a full soccer match and give me an in-depth analysis of team tactics and formations.

Here’s what I need the program to do:
• Map every formation change, pressing trigger, transition from attack to defense, and set-piece routine, highlighting recurring patterns for both our squad and the opponent.
• Track individual player movement and physical output so I can see the exact minute their stamina or football efficiency dips.
• Flag decision-making tendencies—when we choose a long ball, when we overload a flank, how we handle penalties, and similar moments.
• Detect coaching fingerprints by studying substitution timing, positional tweaks, and overall physical game plans, then generate a “profile” of that coach’s style.
• Run a what-if module that projects how a chosen coach (for example, Guardiola or Mourinho) would likely set up for the same match and how that would alter the flow.

Technical expectations
• Accept standard video or event-data feeds (Opta, Wyscout, or comparable tracking data).
• Output interactive visual dashboards plus an exportable report (PDF or JSON) for each match.
• Modular codebase so I can add new coaches or metrics without a full rewrite.

Acceptance criteria
1. Accuracy of formation recognition ≥ 95 % on test footage.
2. Stamina decline point detected within ±2 minutes of manual benchmark.
3. “What-if” coaching module returns tactical differences and predicted xG shift.

When you send your proposal, emphasize your relevant experience in soccer analytics, computer vision, or data science. Links to previous work that shows similar tactical or player-tracking insight will help me judge fit quickly.

Skills: PHP, Java, Python, Software Architecture, Analytics, Software Development, Data Science, Computer Vision

We are looking for a skilled and experienced Time Series Forecasting expert to create accurate predictive models that can forecast future data points. Whether it’s predicting sales trends, weather patterns, or traffic volumes, we need a professional who can help us analyze time-dependent data and provide actionable insights.

(New Zealand & EU/UK)

About the Foundation

We are a newly established charitable foundation operating at the frontier of translational research and precision care. We believe that advances in technology, genomics, data science, and machine learning will transform how we treat patients, empower individuals to take greater control of their health, and unlock healthier living for all. Our mission is to support the discovery of novel care pathways that intervene before patients become sick, shifting the paradigm from reactive treatment to proactive, evidence-based prevention. We provide funding and partner with the world's leading hospitals, physician-scientists, and researchers to explore and apply new approaches to tackling health challenges that go beyond diseases, disciplines, and datasets.

About the Work

We need someone to help us identify and compile a curated list of leading clinician-scientists and academics based in New Zealand and the EU/UK who are active in our focus areas. The list will be used by our team to initiate outreach and explore research partnership opportunities. You are encouraged to use any tools available to you, including AI tools, to complete this task efficiently and to a high standard.

Key Deliverable

A single structured document (preferably Excel) of ~50 individuals across New Zealand and the EU/UK, covering one or more of the following specialisms:

• Iron overload and iron metabolism disorders
• Haptoglobin phenotypes, specifically Hp2-2, and related clinical implications
• Gut health and microbiome research
• Metabolic health

For each individual, please include:

• Full name and title
• Institution and country
• Area of specialism
• Brief note on their research focus (2–3 sentences)
• Link to institutional profile or a key publication
• Any additional information that supports their standing as a leading expert in their field, including, but not limited to, citation metrics, major grants, leadership roles, or notable publications

All entries must be verifiable via institutional websites, PubMed, or Google Scholar.

What We'd Like to See in Your Proposal

• Brief note on how you would approach identifying experts across these geographies and fields
• Any relevant experience in scientific research or biomedical sourcing
• Estimated timeline to deliver the list

see attacProject Title: BI Engineer & Strategic AI Consultant (Looker Studio + Xero + Nookal)
Project Overview:
We are a growing health clinic looking for a Data Engineering and Business Intelligence (BI) expert. Our ultimate business goal is to achieve a $4M business valuation by 2035, which we define as hitting a consistent $5M Annual Recurring Revenue (ARR) with a 20% net profit margin.
To get there, we are automating our monthly reporting and dashboarding. We need a professional to architect and build a data pipeline that pulls clinical data from Nookal (our Practice Management Software) and financial data from Xero, and visualizes it in an interactive Google Looker Studio dashboard.
Crucially, we want to run this combined data through an AI model (OpenAI/Claude) not just to summarize the data, but to act as a Strategic Advisor—suggesting specific actions based on the data to keep us on track for our 2035 valuation goal.
I am looking for a "Build + Handover" package. I need you to architect the pipeline, build the dashboards, configure the AI prompt, and then do a Zoom handover to teach me how to maintain it.
The Tech Stack:
Data Sources: Nookal REST API & Xero API
Data Pipeline / Middleware: Open to your expert recommendation (e.g., Make.com, n8n, custom Python scripts, Coupler.io, Fivetran, etc.). Please propose the most reliable and cost-effective solution.
Database: Google Sheets or Google BigQuery (Happy to take your recommendation based on data volume and longitudinal tracking needs).
Visualization: Google Looker Studio
AI Agent: OpenAI (ChatGPT) or Anthropic (Claude) API
Core Dashboards & Metrics to Build:
The Looker Studio dashboard must be built to allow for both cross-sectional views (e.g., comparing practitioners, locations, or service lines side-by-side for a specific month) and longitudinal views (e.g., tracking revenue, margins, and retention over the next 10 years).
We need the dashboard broken down into the following visual pages/sections:
1. Valuation Goal Tracking (The "North Star" Page)
A pacing dashboard showing current trailing 12-month revenue vs. the $5M goal.
Current net profit margin vs. the 20% goal.
Visualizations: Gauge charts for the overarching goals, and longitudinal trend lines plotting our trajectory to 2035.
2. Deep Financial Health (Nookal + Xero Combined)
Practitioner Profit Margins: Revenue generated per practitioner (Nookal) minus their wages/super/costs (Xero).
Total Payroll-to-Revenue Ratio: Gross wages divided by gross revenue.
True Marketing ROI & Cost of Patient Acquisition (CPA) calculated by matching Xero marketing spend against Nookal new patient numbers.
Average Visit Value (AVV) and Lifetime Client Value (LTV).
3. Capacity & Efficiency
Practitioner utilisation and clinic occupancy (Total occupied hours vs. availability).
Occupied billable vs. occupied non-billable hours.
Leave tracking influence (Pulling leave from Xero and correlating it with Nookal capacity).
4. Retention, Volume & Growth
Consult numbers, new patient flow, and new case numbers per location.
Rebooking rates, Cancellation rates, and DNA (Did Not Attend) rates.
Waitlist Conversion Rate (Percentage of cancelled slots successfully filled).
The AI Component (Strategic Advisory):
Within the data pipeline, an AI step must analyze the raw Nookal/Xero data for the month before it hits the dashboard.
The Prompt Engineering: The AI must be prompted with our goal ($5M ARR / 20% margin). It must analyze the monthly data against this target.
The Output: A 3-4 paragraph advisory text. It should identify what metrics are dragging down our profit margins or growth (e.g., "Practitioner utilization dropped to 65% this month, costing $X in lost potential revenue"), and offer actionable interventions (e.g., "Focus front-desk training on waitlist conversion for the Tuggerah site to recover lost cancellation revenue").
This text summary should auto-populate into a designated text box on the Looker Studio dashboard.
Key Deliverables:
Automated Data Pipeline: A secure, scheduled workflow fetching data from Nookal and Xero APIs and pushing it to a central database.
Looker Studio Dashboard: Build clean, interactive visual dashboards mapped to the metrics above, with date-range filters (for longitudinal tracking) and site-specific filters (for cross-sectional comparisons).
AI Prompt Configuration: Set up the AI module with the advanced prompt engineering required to act as a strategic advisor.
Handover & Training: A 60-minute recorded Zoom session walking me through the architecture you chose, how to troubleshoot pipeline breaks, and how to tweak the AI prompt over time.
Ideal Candidate:
Strong Data Engineering background with experience building API integrations connecting CRM/Clinical software with financial software (Xero/Quickbooks).
Advanced skills in Looker Studio (data blending, time-series analysis, UI/UX design).
High-level understanding of prompt engineering for business strategy.
Fluent English and excellent communication skills for the technical handover.
To Apply:
Please start your proposal by answering these questions:
What pipeline/middleware tool do you recommend for connecting the Xero and Nookal APIs securely, and why?
Would you recommend using Google Sheets or Google BigQuery as the database for longitudinal tracking up to 2035, and why?
Please provide 2-3 links or screenshots of Looker Studio dashboards or complex API pipelines you have built.
hed description for detailed pro

We are seeking a skilled freelancer to design and deliver comprehensive training courses covering AI & Machine Learning, Data Science & Data Analytics, and Full Stack Development. The ideal candidate will have experience in creating engaging content and a strong understanding of these subjects. You'll work closely with our team to tailor the curriculum to our audience's needs. If you are passionate about teaching and have a proven track record in these fields, we would love to hear from you!

Looking for someone who is really excellent at machine learning and data science and other related jobs to help create content at Reddit and a bunch of other technical places, all the time, answer questions like creating content on Quora and other places so that we can get a lot of backlinks, help with SEO and AEO, all of that. You really need to be technical; you need to be a machine learning engineer or a database data scientist or someone, because our product is like that; it is a machine learning model.

Seeking a digital forensics and data analyst to urgently analyze a Google Takeout file for security threats. The ideal candidate will have experience in digital forensics and data analysis, with a focus on identifying potential security risks. This project requires a quick turnaround, so the ability to work efficiently under pressure is essential.

We're looking for a data-savvy person to help us produce monthly reports on the early-career job market. Think hiring trends, student outcomes, salary benchmarks, and employer activity across industries.

What you'll do:
You'll query our SQL database, pull hiring and outcome data, and turn it into clean, compelling monthly reports. The data covers job postings, student career outcomes, employer hiring patterns, and market trends across 1,600+ university programs and 300K+ employers.
Some months the focus will be broad market trends. Other months we'll zoom into verticals like financial services, law, or government. You'll work closely with our marketing team to make sure the data tells a clear story.

What we're looking for:
A degree in data science, statistics, economics, or a related quantitative field. Strong SQL skills are essential since you'll be working directly with our database. Familiarity with AI coding tools (Claude Code or similar) is a plus. You should be comfortable building visualizations and writing short data narratives that a non-technical audience can understand.

Nice to have:
Prior work creating recurring data publications or reports.

Details:
Approximately 15 hours per month. Flexible schedule. You'll coordinate with our team async and deliver polished reports on a monthly cadence. 

We are seeking an expert in AI/ML and Statistics to enhance our securities trading capabilities. The ideal candidate will have a strong background in finance and trading, with the ability to develop and implement advanced trading strategies using machine learning algorithms. Experience with data analysis and statistical modeling is essential. The role involves collaborating with our team to identify market trends and optimize trading performance.

• Describe your recent experience with similar projects
• What techniques would you use to clean a data set?

The Mission
To dismantle the "ROAS Trap." You will be responsible for validating the long-term commercial impact of our branded storytelling. You’ll lead the transition from chasing short-term "clicks" to building "Fame," reducing price sensitivity, and securing market share through sophisticated econometric modeling and regional correlation.

Key Responsibilities
Design the "Balanced Scorecard": Move us beyond Meta’s dashboard. You will build a reporting suite that weights Leading Indicators (Likeability, Brand Linkage, Search Lift) against Lagging Indicators (Value Share, Price Elasticity, NPV).

Orchestrate Regional Experiments: Forget binary On/Off tests. You will manage "Natural Variation" models—analyzing how minor fluctuations in media weight across regions correlate with sales ripples.

Protect the "Brand Tail": Use Discounted Cash Flow (DCF) analysis to calculate the Net Present Value of our campaigns, ensuring we don’t kill high-performing emotional assets just because they don't have a 7-day click attribution.

Hierarchy of Objectives: Work with the C-suite to ensure every campaign follows the IPA-proven hierarchy: Profit, Market Share, Penetration.

Creative Filter: Run "Vibe Checks" on long-form creative before it hits the market, testing specifically for Brand Linkage and Fame Potential to avoid the "funny ad, wrong brand" syndrome.

The Toolkit (What they need to know)
The "Science": Deep experience in Econometrics or Marketing Mix Modeling (MMM). They should know how to "noise-cancel" market data to find the incremental lift.

The "Art": A gut for emotional storytelling. They understand why "Likeability" is a better sales predictor than "Standout."

The "Math": Proficiency in NPV and Marginal Contribution. They shouldn't be confused by "fully loaded" net profit versus the actual margin on incremental sales.

The "Speed": Capability to build 30-day "Proxy Models" so the business can pivot fast without waiting 12 months for a full econometric read.

• How have you implemented Les Binet's work on marketing effectiveness for other clients in the past?
• Let's say we launch 5 ads in one go. How do you validate the long-term commercial viability of those 6–12 month payback branded ad using 30-day digital proxies, and what strategic guardrails prevent the 'ROAS Trap' or the 'Funny Ad, Wrong Brand' syndrome?
• For branded, long-term ads, how would you track the results differently based on placement? TV vs Social Media

Description
We are professors and scholars at Carnegie Mellon University (https://www.cmu.edu). We are seeking a data science freelancer for an academic research project with high impact. The data science freelancer is expected to start work immediately.

Responsibilities
The data science freelancer will be involved in the core stage of the research project. The most important task for the data science freelancer will be to use advanced techniques (including rule-based/modern machine learning/natural language processing/large language model techniques) to analyze consumer behavior from large (hundreds of TBs) web Uniform Resource Locator (URL) datasets of URLs visited by online users. Specific tasks include:
1.  Use information in the URL itself to classify each URL into the category indicating purchase activity and the category indicating ad click activity;
2.  For each user in the dataset, estimate each individual’s total number of purchase activities and ad click activities over a period, based on the information from Step 1.
The data science freelancer is expected to update their progress on a daily basis and meet regularly with the research team.
At the end of the project, the data science freelancer is expected to deliver:
1. High-quality solution + code
2. Well-written technical document

Qualifications
Successful candidates will have demonstrated relevant expertise and experience, including:
a.  Expertise in web Uniform Resource Locator (URL) analysis
b.  Strong programming skills in Python, Spark, and SQL
c.  Experience working with large (TB-sized), complex datasets
d.  Expertise in using modern machine learning/natural language processing/large language model techniques
e.  Experience working with Linux/Unix-based systems
f.  Strong written and oral communication skills in English
g.  Independent judgement, organization, attention to detail, and patience

Hey,

I have a Pinescript version 6 strategy in Tradingview that needs to be converted one on one for Ninja Trader 8. It has to be exactly the same. Because I ll need to use it to automate my accounts. Wondering if its possible and what ll be the cost? Also FYI no AI models were able to replicate it for me. I know basic coding skills and these sites couldnt do this.

Thank You
Amaninder

I need data for my industry downloaded via api ( i think) & then analyzed. I need someone very experienced in python to do this.
20 is hours a week. English must be excllent, excellent work ethic, attention to detail is a must, etc. Looking for a real go getter who knows python well & has good english who isnt afraid to get to work!.

I have a Python-based data scraper that is currently running too slow and crashing due to high memory usage when processing large datasets. It’s built with Selenium/Requests but isn't scalable.
Requirements:
-Refactor sync loops to Asyncio/Aiohttp.
-Optimize memory footprint for 5,000+ records.
-Implement a robust error-handling and retry mechanism.
Need an expert who can handle high-performance Python code. No beginners please.

Skills: Python, Data Processing, Web Scraping, Machine Learning (ML), Software Development, Data Science, Web Development, API Development, Performance Tuning

I need a system or service to verify handwritten signatures. The signatures to be verified will be provided as scanned images and will be checked against previous submissions.

Requirements:
- Verify signatures from scanned images
- Compare against a database of previous submissions
- Accuracy and reliability are critical

Ideal Skills and Experience:
- Experience in image processing and signature verification
- Familiarity with machine learning algorithms for pattern recognition
- Attention to detail and high accuracy in verification tasks
- Ability to deliver results within the project timeline

Please provide examples of similar work done and your approach to handling this task.

Skills: Machine Learning (ML), Data Science, Neural Networks, Image Processing, Data Analysis, Computer Vision, Deep Learning, Image Analysis

I’m scaling the Data Science division at Finance King Technologies and have 90 remote openings for AI / ML Data Model Validation & Research. Your core mission is to push model accuracy higher on our live predictive-modelling, natural-language-processing, and image-recognition pipelines.

You will receive full in-house training, yet I do expect you to arrive with:
• Python programming confidence
• Hands-on exposure to common machine-learning algorithms
• A solid statistical-analysis mindset

Whether you are an undergraduate in your final years, a fresh graduate, or already carry up to five years of experience, you are welcome as long as you are 18 + and located anywhere in India. The position is permanent, pays ₹8,00,000 per annum, and is 100 % remote.

Day to day, you’ll design validation experiments, audit datasets, benchmark model iterations, and communicate findings that guide the engineering teams. Expect clear deliverables—structured validation reports, annotated notebooks, and brief walkthrough sessions—to mark each milestone.

If rigorous analysis, constant experimentation, and the chance to see your work improve real-world financial products excite you, let’s talk.

Location: 100% remote within India region only

Skills: Finite Element Analysis, Machine Learning (ML), Statistical Analysis, SPSS Statistics, Data Science, Data Visualization, Data Analysis, Natural Language Processing, Machine Learning Algorithms, Image Recognition

We are an ecommerce brand using Triple Whale and looking for an expert who can turn our data into clear, actionable dashboards and reports. You should be very comfortable with Triple Whale’s SQL editor (ClickHouse-flavored SQL), data modeling for analytics, and building visualizations that answer real business questions (ROAS, MER, LTV, cohort performance, channel attribution).

• Describe your experience using Triple Whale’s SQL editor. What kinds of reports or dashboards have you built in Triple Whale specifically
• Triple Whale can combine multiple data sources (paid media, Shopify, etc.). How would you validate that a new SQL-based report is correct and aligned with the brand’s business logic before sharing it with stakeholders?
• We are newly onboarded to Triple Whale and do not have much historical data. How would you design SQL queries and reports that account for limited history (for example, hardcoding dates, channels, or other parameters) so the numbers are still reliable
• Describe your recent experience with similar projects

*PLEASE, no auto reply. I won’t listen, start your msg with telling me in what way an LLM can give honest critique about an image and help them improve dramatically. What key features should a report have that can help influence them*

Description:
I’m building a creative feedback tool for filmmakers and visual artists. I need a developer to build the backend pipeline that takes an uploaded image or video file (think short form content, ideally larger files if possible to be able to review short films) and generates a structured critique report combining two AI layers. Upon report generations, users will have the option to have additional feedback from a human if they choose too.

Layer 1 — Neural Attention Analysis (Meta TRIBE v2)

Here’s a reference video:

https://x.com/chrisclark/status/2039353858636271758?s=46
As you can see in this video, his report is able
To generate details about focus, brain activity etc.
Meta recently open-sourced TRIBE v2, a foundation model that predicts how the human brain processes visual stimuli. It creates a digital twin of neural activity from fMRI data. I need this model integrated to analyze uploaded images and return:
∙ Predicted attention pathways — where the brain looks first, second, third
∙ Visual cortex activation intensity — how much cognitive effort the image demands
∙ Attentional engagement score — does the image capture and hold attention or does the brain scan and move on
∙ Sensory processing summary — is the brain processing this image efficiently or fighting the composition
The model weights and codebase are publicly available under CC BY-NC here: https://github.com/facebookresearch/TRIBE (or search “Meta TRIBE v2 foundation model”). The interactive demo is at https://aidemos.atmeta.com/tribev2. The model processes images through a vision transformer and predicts whole-brain fMRI responses.
I don’t need a full neuroscience research output. I need the model’s predictions translated into plain-language insight that a filmmaker or designer can actually use. For example: “The brain processes the center of this frame first, then moves to the upper left. The lower right receives minimal attention — if there’s important information there, it’s being missed.”

Layer 2 — Art Principles Critique (Claude Vision or GPT-4o)
A vision-capable LLM analyzes the same image and generates structured scoring and feedback across these dimensions:
∙ Contrast (value separation, visual pop)
∙ Balance (weight distribution, intentional asymmetry)
∙ Emphasis (focal point clarity)
∙ Movement (visual pathways through the frame)
∙ Unity (cohesion of elements)
∙ Proportion (scale relationships)
∙ Color Harmony (palette coherence, temperature, mood)
∙ Craft (technical execution, lighting, finish)
Each dimension gets a 1-10 score and a 1-2 sentence note specific to the image.
Additionally, the LLM should return:
∙ A color palette extraction (3-9 dominant hex codes, a style name, and effectiveness note)
∙ A creative archetype classification (one of 8 predefined types based on the visual choices present)
∙ A brief summary of what the image is doing and a one-line verdict
∙ A “Director’s Note” — the single most important change
∙ A “Steal This Move” — the strongest element and why it works
Combined Output
Both layers feed into a single structured JSON response that I can render as a branded report on the frontend. The neural analysis and the art critique should reference each other where relevant — for example if TRIBE v2 predicts low attention in an area, and the art critique scores Movement as weak, those should be connected in the output.
For video submissions: extract 3-5 key frames at intervals and run both analyses per frame. Return a summary that covers the video as a whole plus frame-level breakdowns.
Tech requirements:
∙ TRIBE v2 model running on GPU inference (the model requires CUDA)
∙ Claude API or OpenAI API for the vision critique layer
∙ Input: image file (JPG, PNG, WEBP) or video file (MP4, MOV) via URL or upload
∙ Output: structured JSON matching a schema I’ll provide
∙ Deployed as an API endpoint I can call from my frontend (hosted on Vercel/Supabase)
Ideal candidate:
∙ Experience running open-source ML models (PyTorch, transformers) on GPU infrastructure
∙ Familiar with Meta FAIR research repos or similar neuroscience/vision models
∙ Experience with vision LLM APIs (Claude, GPT-4o)
∙ Can deploy and host a GPU inference endpoint (Lambda Labs, RunPod, Modal, or similar)
∙ Bonus: experience with fMRI data interpretation or computational neuroscience
Budget: Open to proposals. This is an MVP build — I need a working pipeline, not a production-scale system. Happy to discuss scope.
To apply: Tell me (1) your experience running open-source vision/neuro models, (2) how you’d architect the two-layer pipeline, and (3) a rough estimate on timeline and GPU hosting costs.

That’s ready to paste. The post is specific enough that unqualified people won’t apply and qualified people will know exactly what you need. The “to apply” filter at the bottom saves you from wading through generic proposals.​​​​​​​​​​​​​​​​