Akshay Patel
Data Scientist / ML Engineer · Boston, MA · Open to ML, Data Science, and MLOps roles
Always curious, always learning.
I build end-to-end ML systems: statistically rigorous analysis on the front end, production pipelines on the back end. My background is an M.S. in Industrial Engineering (WVU), so I care about the things that make models trustworthy in the real world: leakage-free validation, survey-weighted inference, causal identification, drift monitoring, and reproducible pipelines.
📧 akshaypatelnitb6@gmail.com · LinkedIn · GitHub · About
Featured Work
| Project | What it shows | Stack |
|---|---|---|
| arXiv RAG | Hybrid BM25 + vector retrieval over daily-ingested papers, multi-turn chat, local LLM | FastAPI, OpenSearch, Ollama, Airflow, Redis, Langfuse |
| ML Monitoring & Drift Detection | PSI drift detection, champion/challenger retraining, model registry | PySpark MLlib, MLflow, Prefect, Prometheus, FastAPI |
| Demand Forecasting at Scale | 74M-row forecasting with strict temporal isolation and quantile intervals | LightGBM, MLflow, SHAP, FastAPI |
| Pollution & Cancer (BRFSS) | Survey-weighted inference + ML on 433K respondents | samplics, XGBoost, SHAP |
| Causal Inference Toolkit | PSM, IPTW, doubly-robust estimators with sensitivity analysis | EconML, DoWhy, statsmodels |
| Supply Chain Agents | Multi-agent LLM system with supervisor routing and guardrails | LangGraph, LangChain, Streamlit |
Skills
- ML & Statistics: supervised learning (GBMs, linear models), time-series forecasting, causal inference (PSM, IPTW, AIPW/DML), A/B testing, survey statistics, PCA/clustering, SHAP interpretability
- LLMs & Retrieval: RAG, hybrid search (BM25 + k-NN + RRF), multi-agent orchestration (LangGraph), guardrails, LLM observability
- MLOps & Data Engineering: MLflow, Airflow, Prefect, Kestra, dbt, PySpark, Docker, FastAPI, drift monitoring, Terraform
- Languages & Data: Python, SQL, C++, JavaScript · PostgreSQL, BigQuery, DuckDB, MongoDB, OpenSearch, Redis
- Cloud: GCP (BigQuery, Cloud Run, GCS), AWS
Experience (short version)
- Researcher / Data Analyst, Grant Fund Implementation — Data Driven WV · Jul 2026 – Present Own the data science lifecycle for two state grant programs, from clustering and classification models to Tableau and ArcGIS tools that show where funding should go.
- Cybersecurity Analyst — Data Driven WV · Aug 2025 – Dec 2025 Led a team of 4 building an NLP decision-support system for NIST 800-53 compliance (MiniLM embeddings with rule-based validation), which cut manual audit processing time by 30%.
- Graduate Research Assistant — WVU IMSE Department · Jan 2024 – Dec 2025 Led data-driven energy audits for 9 facilities, finding 1M+ in USDA REAP grants.
Full details on the About page.
Also here
| Section | Description |
|---|---|
| ◎ Learning | TIL notes, snippets, and what I’m reading |
| ◈ Blogs | Long-form essays and deep dives, like System Design Basics |
| ◉ Off-Grid | Hobbies and the person behind the code |