Hello! I'm
SUR
VAGHASIYA
A
Software Engineer, AI/ML Engineer, and Data Scientist
AI/ML · Data Science
About Me
I started out building websites, got curious about what was happening behind them, and never quite stopped pulling that thread. It led me to teach a computer to count cars, ship a forecasting service that real teams relied on, and map where crashes keep happening.
Then came the AI rabbit hole: an assistant that reads your documents, an analytics app you can argue with in plain English, a hackathon voice agent that checks in on rare-disease patients, and FRIDAY, the voice assistant on my Mac, which by house rule isn't allowed to claim anything it didn't do. Somewhere in there, I also wrote a smart contract that trusts no one. On purpose.
So: software engineer, AI engineer or data scientist? I've stopped trying to pick. Have a look around, and if anything sparks an idea, come say hi.
WHAT
I DO
SOFTWARE ENGINEERING
Backend services, APIs & native apps
FRIDAY: a Swift macOS app running three Go services over typed IPC, with 1,666 automated tests. Plus REST and WebSocket APIs (FastAPI, Flask), PostgreSQL, and Dockerized stacks.
Skillset & tools
FULL-STACK & CLOUD
Web products, deployment & CI/CD
End-to-end Next.js + TypeScript products — SAARTHI's doctor dashboard, InsightX's real-time app — shipped on Vercel, Railway, and AWS with CI/CD, plus a Web3 escrow on Cloudflare.
Skillset & tools
AI, ML & DATA SCIENCE
LLMs, RAG, voice AI & forecasting
Voice AI with Claude (SAARTHI, HackRare 2026), multimodal RAG with 16 agentic tools (EchoMindAI), ensemble forecasting (+33%), and geospatial ML on 33K crash records.
Skillset & tools
My career &
experience
Data Science & AI Intern
Adani Enterprises Ltd.
2025
Jan – Jul 2025 · Ahmedabad, India
Shipped a production sales & revenue forecasting service (Python, Flask, Scikit-learn; 78% accuracy) on AWS — EC2 for serving, S3 for artifacts, Lambda for scheduled retraining — with GitHub Actions CI/CD that cut deployment time 44%. Dashboards replaced manual reporting for 10+ product teams, saving 34% of analysis time; automated retraining cut operational overhead 37%.
Python Machine Learning Intern
Techmicra Data Systems
2024
Jun – Jul 2024 · Ahmedabad, India
Built Git-versioned ML pipelines in a five-member Agile team, plus real-time vehicle detection and traffic counting on live video with YOLO. EDA (ANOVA, correlation matrices) boosted model interpretability 42%; Grid Search CV cut overfitting 35% and improved generalization 27%.
Web Developer Intern
Shaligram Infotech
2023
Jun – Jul 2023 · Ahmedabad, India
Built and shipped 5 responsive cross-browser web pages (HTML, CSS, JavaScript) with 48% faster load times, and 2 MySQL-backed WordPress sites for real clients — improving delivery speed and client satisfaction by 45%.
Education
MS, Data Science
Stony Brook University · New York
2027
Aug 2025 – May 2027 Coursework: ML, Big Data Systems, Big Data Analysis, Algorithms & Networks, Data Structures & Algorithms, Statistical Computing, Probability & Statistics, Data Analysis & Management, CS Fundamentals. Led research on LI Traffic Accident Analytics — 33K+ records, DBSCAN, XGBoost (AUC-ROC 0.72), SHAP, GeoPandas, NetworkX.
BTech, Computer Science & Engineering
Indus University · Ahmedabad, India
2025
Sep 2021 – May 2025 · GPA 3.4/4.0 Coursework: ML, NLP, Deep Learning, DSA, DBMS, Cloud Computing, Web Technology, Software Engineering. Covered the full CS spectrum — from algorithms & systems to ML, cloud, and applied engineering.
Honors & Awards
2026
Built SAARTHI — AI voice clinical monitoring platform for rare-disease symptom tracking. Combined Retell AI phone calls, Claude transcript analysis, FastAPI/Flask, PostgreSQL, personalized ML models, and a Next.js doctor dashboard.
My Work
01
FRIDAY
Systems / macOSA macOS voice assistant I use every day — a first step toward AI that keeps context across apps and never claims an action it didn't take. A Swift menu-bar app supervises three Go services over typed IPC: the LLM only proposes, local code verifies. On-device wake word, three-tier voice fallback, self-healing restarts, and 1,666 automated tests.
Tools and features
Swift · Go · Unix-socket IPC · SQLite · sherpa-onnx · Cartesia TTS · OpenAI · Groq
02
LI Traffic Analytics
Data ScienceWhere and why severe crashes happen on Long Island: 12 hotspot clusters (DBSCAN) and 4 risk profiles (FAMD + k-Means) from 33,228 accident records, an XGBoost severity model (AUC 0.72) explained with SHAP, and road-network overlays (GeoPandas, osmnx) for safer routing.
Tools and features
Python · R · XGBoost · DBSCAN · SHAP · GeoPandas · osmnx · NetworkX · scikit-learn
03
SAARTHI
Full-Stack / Voice AIHelps Empty Nose Syndrome patients track symptoms between doctor visits: an AI voice agent calls daily and turns how they feel into trends doctors can act on. Co-built in 24 hours with a team of 4 at HackRare 2026 — I owned the FastAPI/Flask REST APIs, the PostgreSQL schema, and the Next.js doctor dashboard that streams Claude replies in real time.
Tools and features
Next.js · TypeScript · FastAPI · Flask · PostgreSQL · SQLAlchemy · Claude · Retell AI · Vercel · Railway
04
InsightX
Full-Stack / MLA real-time forecasting platform I built to learn how ML supports business decisions — upload data, get forecasts and plain-English answers. Seven Dockerized services (FastAPI, Next.js, PostgreSQL, Redis, MinIO, MLflow, MCP); Redis Pub/Sub + WebSockets stream 5–30 s forecasts with zero timeouts; an ensemble model lifts accuracy 33%.
Tools and features
FastAPI · Next.js · TypeScript · Redis Pub/Sub · WebSockets · PostgreSQL · Alembic · Docker Compose · MLflow
05
EchoMindAI
AI / RAGAn AI assistant that answers from your own files (PDF, Word, slides, spreadsheets) and live web data, by text or voice. A LangChain agent with 16 live tools over FAISS vector search; moving speech-to-text to Groq cut latency 7.5× (1.5 s → 200 ms); layered error recovery cut failures 90% with zero crashes.
Tools and features
Python · LangChain · FAISS · GPT-4o · Groq · Streamlit · Docker · MCP
06
Decentralized Escrow
Web3 / ServerlessTrustless payments without a middleman: a Solidity contract on Ethereum Sepolia locks funds until both parties approve, with a refund path so money never gets stuck. Serverless Hono API on Cloudflare Workers with Upstash Redis, a React + MetaMask frontend, and Hardhat tests for every contract state.
Tools and features
Solidity · Hardhat · ethers.js · React · Vite · Hono · Cloudflare Workers · Upstash Redis
Tech Stack
Let's build something
extraordinary.
Get In Touch.
Send a message
Availability
Open to full-time roles
© 2026 Sur Vaghasiya