Hitesh Kalwani

Built a ChatOps AI assistant integrated with Kubernetes and CI/CD pipelines, enabling natural language based deployments, log retrieval, and system diagnostics. Reduced manual DevOps effort by60% using LLM function-calling and automated workflows. Developed an internal AI engineering copilot using FAISS, LangChain, and Model Context Protocol (MCP), enabling contextual querying across distributed system documentation, runbooks, and codebases. Processed large-scale document corpora for semantic retrieval across engineering knowledge bases, implemented document chunking, embedding pipelines and hybrid retrieval (semantic + keyword), improving answer accuracy by35% and reducing onboarding time by ~50%. Designed a real-time distributed system connecting2M+ users to network speakers using WebSocket, WebRTC, and async Python. Achieved high-throughput, low-latency communication using Kubernetes, Redis, and distributed architecture. Built a high-scale job scheduling system handling10,000+ jobs/sec using Apache Pulsar and Node.js, improving system scalability by30%. Led sensor integration platform supporting100,000+ devices, enabling real-time monitoring and improved decision-making.

Built a near real-time analytics platform ingesting4M+ events/day using AWS Kinesis, Lambda, and Redshift. Designed serverless data pipelines powering dashboards for10M+ users across multiple regions. Developed a Computer Vision + NLP-based cost optimization system (Mask RCNN, OCR), reducing manual processing and improving workflow efficiency. Improved API performance by60% using optimized microservices architecture on AWS ECS, Lambda, AWS Step Functions and HTTP API with API Gateway. Led AI team across full SDLC, including system design, CI/CD implementation, and delivery of scalable ML solutions.

Built a receipt entity recognition system using Python, Django, REST APIs, AWS, Sagemaker, NLP, OCR, and deep learning (PyTorch), reducing processing time by50%+. Developed a driver safety prediction system using real-time data (weather, vehicle health), achieving 89.93% accuracy. Delivered ML pipelines integrated into production systems for large-scale enterprise use cases.

Built high-volume insurance systems processing1M+ requests/day and managing1B+ records. Developed analytics tools processing10M+ records using .NET, SQL Server, and ETL pipelines. Designed REST APIs and workflows for enterprise systems, improving delivery efficiency and system scalability. Secured 2nd Rank in internal hackathon for building a traffic signal recognition system using OpenCV, NVIDIA Digits, Python, Machine Learning and Deep Learning.

Contributed to distributed systems supporting3M+ devices using Node.js, ElasticSearch, Angular and CouchDB. Built end-to-end testing pipelines improving system reliability and deployment quality.