milan = {
"location" : "NYC Metropolitan Area π½ (Open to Relocation)",
"education" : "M.S. Computer Science, Stevens Institute of Technology (GPA: 3.97 Β· Dean's List)",
"experience" : "3+ years production AI/ML @ Tata Consultancy Services",
"focus" : ["Agentic AI", "LLM Infra", "MLOps", "Distributed Systems"],
"currently" : "Teaching Assistant β Algorithms & Statistical ML @ Stevens",
"seeking" : "Full-time roles in AI/ML Engineering, Agentic Systems, MLOps",
}π Teaching Assistant β Algorithms & Statistical ML Β· Stevens Institute of Technology, NJ
Jan 2025 β Present
Mentor 130+ students across CS 590 (Algorithms: DP, graphs, backtracking) and CS 560 (Statistical ML: PAC learning, VC dimensions, Transformers). Automated grading pipelines delivering feedback 3Γ faster.
π Graduate Assistant Β· Graduate Education Office, Stevens Institute of Technology, NJ
Oct 2024 β Dec 2024
Compiled and reported career outcomes of 1,000+ students via calls, emails, and survey forms. Automated data collection using MS Excel and the department's reporting application API β increasing results by 3Γ.
π€ AI/ML Engineer Β· Tata Consultancy Services Ltd., Mumbai, India
Jul 2021 β Aug 2024
Architected a production agentic chatbot (GPT-4, LangChain, Azure Cognitive Search) processing heterogeneous docs at scale + Context-Driven Search API boosting relevance by 200%. Engineered real-time KafkaβMongoDB pipelines serving 10 global regions Β· 300+ client apps. On-call optimization preventing $1M+ revenue loss.
π¨βπ« Programming Teacher Β· Neebal Learning Pvt. Ltd., Mumbai, India
Jun 2019 β Jun 2021
Taught 100+ engineering students in Python, Java, C, C++, DBMS, and Data Structures & Algorithms β emphasizing clean, robust, bug-free programming practices.
| π₯ | 1st Place β Chubb Insurance AI Challenge, QuackHacks '26 |
| π | Dean's List all semesters Β· GPA 3.97/4.0 |
| π° | $1M+ revenue impact via on-call production optimization @ TCS |
| π | 200% improvement in enterprise search relevance (vector DB + cosine similarity) |
| π€ | 15+ ML models deployed at edge on Raspberry Pi 4 (90%+ macro accuracy) |
| π | Real-time pipelines serving 10 global regions Β· 300+ client applications |
Languages
AI / LLM / Agentic
ML Frameworks
Cloud & Infra
Data Engineering
Full-stack multimodal agentic system β React frontend (webcam/audio) + FastAPI backend orchestrating a Google ADK agent via Gemini Live API with bi-directional WebSocket streaming. Deployed on Google Cloud Run.
Google ADK Gemini Live API FastAPI React WebSocket Cloud Run Docker
AI-powered fraud detection across 5 insurance verticals using a dual-layer forensic pipeline β parallel CV forensics (ELA, FFT, Noise, JPEG Grid, EXIF) fused with Gemma 3 27B vision inference via weighted majority-vote scoring engine. Batch mode with cross-image MD5 dedup + ELA outlier detection (Z > 2Ο).
Gemma 3 27B FastAPI React Pillow NumPy Vercel Railway
Real-time multivariate time-series forecasting (LSTM/XGBoost/ARIMA) on live GBFS feeds achieving RΒ²β0.96. Geospatial Streamlit dashboard for station-level risk assessment served via FastAPI REST API.
TensorFlow XGBoost ARIMA FastAPI Streamlit REST APIs
Transformer EBM + Hierarchical VAE architecture for latent-space iterative reasoning via MCMC sampling. Diagnosed posterior collapse and energy flatness instabilities over 400+ GPU-hours.
PyTorch Transformers Hierarchical VAEs MCMC EBM
15 ML models (90%+ macro accuracy) deployed on Dockerized Raspberry Pi 4 for UAV intrusion detection using sensor waveform + cyber data. Benchmarked latency, throughput, and compute constraints for embedded real-time inference.
CNN LSTM SVM Random Forest Docker Raspberry Pi 4 Edge AI
Distributed multi-class classifiers on Tahoe-100M genomic dataset using Spark ML + PCA on GCP Databricks. Large-scale biological ML pipeline scalability analysis.
PySpark Spark ML PCA Databricks GCP
I'm actively seeking full-time roles in AI/ML Engineering, Agentic Systems, MLOps, and Data Engineering.
Based in NYC Metro β open to relocation.
