I'm an AI Engineer who builds things that ship, not just prototype.
Founding ML Engineer at two startups where I built complete AI stacks from scratch: OCR pipelines, face recognition systems, LLM fine-tuning, and text-to-speech engines, replacing outsourced APIs with in-house solutions serving 500K+ requests/month.
Currently building autonomous multi-agent systems with LangGraph, MCP, and A2A protocols. MS in Computer Engineering (CV & ML) from Northeastern University, graduating Dec 2025.
I care about AI systems that actually work in production: real monitoring, real drift detection, real CI/CD, not just a notebook that runs once.
LangGraph | MCP | A2A Protocol | Mem0 | DeepSeek R1
5-agent system (Monitor, Diagnose, Policy, Remediate, Report) that autonomously resolves P3/P4 incidents. Built with LangGraph state machines, A2A protocol for cross-agent task delegation, MCP integrations for Prometheus/Kubernetes/GitHub/PagerDuty, and an AI Governance layer with human-in-the-loop approval gates. Optimized costs 87% via intelligent model routing.
LangGraph | RAG | Cohere Reranker | GCP | Datadog
Agentic RAG system with conditional routing that dynamically selects between vector search and web search. Multi-step reasoning engine decomposes complex queries into subtasks (Retrieve, Summarize, Compare). Hybrid search + Cohere reranking achieves 0.89 context precision. Deployed on GCP with CI/CD, MLflow tracking, and Datadog monitoring.
FastAPI | Evidently | Prometheus | Grafana | OpenTelemetry | Prefect
Production-grade observability for ML models: real-time drift detection across 8 drift types, distributed tracing, and Kubernetes-ready deployment. Monitors fraud detection, price prediction, and churn models with synthetic drift injection for testing.
XGBoost | FastAPI | Docker | MLflow | Terraform | Airflow
End-to-end ML microservice: model training with Optuna tuning (0.84 AUC-ROC), FastAPI serving with <100ms P95 latency, Terraform infrastructure, Airflow retraining, and Datadog drift monitoring.
ML Engineer @ New Era Technology (Client: AssuredPartners Insurance)
Built and deployed claims prediction models on AWS EKS (0.87 F1-score). Engineered data pipelines with PySpark/Databricks processing 10M+ records daily. Implemented time series forecasting with Prophet for proactive risk mitigation.
ML Intern (Founding Team) @ Instaread
Led R&D of in-house GenAI solutions replacing external APIs (15x cost reduction). Fine-tuned GPT-3.5 Turbo (BLEU 0.42 → 0.58). Built LLM-as-a-Judge evaluation framework for automated quality assessment.
ML Developer (Founding Team) @ Think360.ai
Solely architected the company's complete AI suite: OCR (PaddleOCR, 92% accuracy), Face Recognition, Liveness Detection, Anti-Spoofing. Deployed on AWS EC2 with Docker Swarm, serving 500K+ API requests/month at 99.9% uptime. Partnered with BFSI clients to deliver scalable eKYC solutions, reducing fraud by 60%.
AI/LLM Engineering:
ML/Deep Learning:
Languages:
Infrastructure & MLOps:
Monitoring & Eval:
Northeastern University, Boston — M.S. in Computer Engineering (CV & ML) | GPA: 3.9/4 | Dec 2025
University of Mumbai, India — B.E. in Electronics & Telecommunication | GPA: 8.4/10 | May 2022
Building AI systems that ship to production, not just run in notebooks.