Open to AI engineering roles

Résumé

Applied AI Engineer · Ahmedabad, India

Mubin Attar

Applied AI Engineer with 3+ years in software engineering and a production-AI focus since 2024. I build LLM applications, agentic and RAG systems, predictive ML, and healthcare automation across FastAPI, Next.js, PostgreSQL, and Docker. Four public AI projects document my work in authentication, CI/CD, validation, and task-level evaluation.

Experience

AI/ML Engineer · Sevina Technologies

Ahmedabad · 2024 – present
  • Build and ship production AI & automation systems for healthcare — end-to-end pipelines (Python, Playwright, Selenium) that extract, validate, and synchronize clinical, resident, and demographic data into SQL Server, with logging, reconciliation, and error recovery.
  • Developed an AI clinical decision-support platform that turns CMS MDS assessments into interactive dashboards, compliance findings, and reimbursement insights (HIPPS/PDPM) for skilled-nursing facilities via automated PDF extraction and deterministic, guideline-aligned rules.
  • Built Generative AI applications — LLM document-analysis assistants with multi-provider routing (OpenAI, Claude, Gemini), RAG, and validated structured output — plus a FastAPI eligibility-verification service with async batch processing, Dockerized deploys, and CI/CD.

Junior Python Developer · Linescripts Software Pvt Ltd

Pune · 2022 – 2024
  • Delivered end-to-end web applications — including a Hospital Management System and e-commerce platforms — with Python, Django & DRF, and responsive HTML5/CSS3/JavaScript/Bootstrap frontends.
  • Built RESTful APIs and backend services, integrated third-party APIs, and collaborated cross-functionally on requirements, design, and testing.

Featured projects — case studies & source code

DBWhisper

livegithub

Natural-language-to-SQL agent — LangGraph pipeline, schema-aware pgvector retrieval, multi-LLM fallback, and a fail-closed read-only safety layer.

101/139 Spider dev · 18/22 golden-set exact · 4/4 unsafe prompts refused · FastAPI · LangGraph · Next.js · method

TradePulse

livegithub

Quant backtesting platform with next-bar execution, costs modeled by default, Decimal accounting, and an AI copilot grounded in the backtest results.

Look-ahead canary + cash-conservation tests green · FastAPI · Next.js · TimescaleDB · Redis · method

CrownWager

livegithub

Sports-betting analytics & ML predictions — +EV best bets, a validated XGBoost model, cross-book arbitrage finder, and a graded, verifiable track record.

65.2% ± 0.8% XGBoost accuracy (5-fold CV) · Django/DRF · FastAPI · Next.js · method

LLM Studio

livegithub

Multi-user, ChatGPT-style AI chat SaaS — Argon2id auth, per-user tenancy, token streaming, OpenAI-compatible multi-LLM routing, and a shared-key daily quota.

Per-user tenancy · 31 tests · live on HF + Neon · FastAPI · Postgres · Docker · CI · method


Education

Master of Computer Applications (MCA)

Savitribai Phule University of Pune

2021 · CGPA 8.0