Senior Software Engineer at Stealth/Unknown (Remote)

Owned backend architecture for client-facing MERN products built on Node.js, Express, and Next.js, sustaining 1.2M+ daily API calls with p95 latency under 220ms via SSR and edge caching/strategies. Embedded AI into existing production systems with RAG-powered search and agentic automation on OpenAI and LangChain using Pinecone and Qdrant, improving answer relevance by 47% and trimming support workload. Standardized LLM agent data access using Model Context Protocol (MCP) connectors spanning 12+ internal and third-party services, replacing ad-hoc scripts and saving operation/support time (~50% lookup time). Decomposed a monolith into event-driven microservices over Kafka and RabbitMQ handling 100k+ jobs daily, with idempotent retries and dead-letter queues that pushed failed-job retries down 60%. Reworked MongoDB data models and compound indexes and layered Redis/ElastiCache, halving slow query times and trimming database load by ~one-third. Built WebSocket live-update channels for dash

Lead Data Scientist Engineer at 47Billion

Led development and deployment of advanced Generative AI, LLM, and Retrieval-Augmented Generation (RAG) systems for enterprise clients across healthcare, finance, and telecom. Designed scalable workflows using LangChain, LangGraph, and vector databases such as Pinecone and FAISS, improving query efficiency by 40–60% and saving 20+ hours/week. Built persistent conversational agents using Llama3.2 and LangSmith, deployed via Azure Kubernetes Service. Implemented end-to-end MLOps pipelines on GCP and Azure for automated training, deployment, and monitoring. Built an LLM evaluation framework using Ragas and custom metrics to reduce hallucinations and improve enterprise grader reliability.

Lead Data Scientist/Engineer at 47Billion

Led development and deployment of advanced Generative AI, LLM, and Retrieval-Augmented Generation (RAG) systems for enterprise clients across healthcare, finance, and telecom. Designed scalable workflows using LangChain, LangGraph, and vector databases (Pinecone, FAISS). Delivered 40-60% improvement in query efficiency and 20+ hours/week time savings through intelligent automation. Built persistent conversational agents using Llama3.2 and LangSmith; deployed via Azure Kubernetes Service. Implemented end-to-end MLOps pipelines on GCP and Azure for automated training, deployment, and monitoring. Partnered with cross-functional teams in Agile to drive scalable AI delivery. Pioneered LLM evaluation frameworks with RAGs and custom metrics to ensure hallucinatio n-free outputs and enterprise-grade reliability.

Senior Software Engineer at Stealth/Unknown (Remote)

Delivered full-stack features for time-tracking and expense-management SaaS using React, Angular, and Node.js/Express APIs serving 120K+ monthly active users. Developed scalable Node.js microservices using Express, standardizing validation, pagination, and error handling to cut average integration time for new clients by 30%. Re-architected synchronous workloads into asynchronous BullMQ and RabbitMQ event pipelines, offloading heavy processing and reducing response times by 30%.

Developed production-ready ML pipelines using Python and GCP VertexAI from experimentation to deployment. Led and mentored a team of data scientists to deliver scalable models under tight timelines. Built graph-based recommendation and churn prediction models using neural graph architectures. Deployed a hierarchical frequently-bought-together model on 4M+ transactions, increasing basket size by 8%. Designed end-to-end data workflows integrating BigQuery, time-series modeling, and optimization engines, and optimized model performance and inference pipelines to improve latency and production scalability. Architected a real-time CLV forecasting system using gradient boosting, improving retention campaign ROI by 22%. Championed MLOps adoption using MLflow for experiment tracking and model versioning, reducing retraining cycles by 35%.

Senior Data Scientist at Minute7

Developed production-ready ML pipelines on Python and GCP Vertex AI from experimentation to deployment. Led and mentored a team of data scientists, ensuring delivery of scalable models under tight timelines. Built graph-based recommendation and churn prediction models using neural graph architectures. Deployed a hierarchical frequently-bought-together model on 4M+ transactions, increasing basket size by 8%. Designed end-to-end data workflows with BigQuery, time-series modeling, and optimization engines. Optimized model performance and inference pipelines for latency and scalability. Architected a real-time customer lifetime value (CLV) forecasting system using gradient boosting, improving retention campaign ROI by 22%. Championed MLOps adoption by implementing MLflow for experiment tracking and model versioning, reducing retraining cycles by 35%.

Senior Software Engineer at Evolve Squads

Implemented real-time WebSocket and Redis pub/sub notifications, replacing polling and reducing server load by 42% while achieving near-instant updates. Tuned MongoDB and PostgreSQL queries and indexes on high-read collections, improving dashboard response times by 45%. Established automated CI/CD pipelines with GitHub Actions and Docker (lint/unit/E2E), cutting deployment time and reducing defects escaping to production by 35%. Raised test coverage to 88% using Jest and Cypress suites, enforcing engineering standards via code reviews and reducing production regressions. Deployed and monitored services on AWS and Heroku using blue-green release strategies, achieving zero-downtime rollouts.

Built an NLP-based event detection pipeline on live Twitter streams. Integrated Twitter Live API for streaming ingestion and applied entity extraction for event clustering. Implemented burst detection algorithms achieving 85% accuracy for real-time event identification. Applied Dynamic LDA for topic modeling with 82% classification accuracy. Designed scalable data pipelines supporting multi-window temporal analysis and high-volume ingestion. Improved model robustness through feature engineering, evaluation metric tuning, and validation workflows. Engineered a Spark-based streaming architecture on GCP Dataflow to process 50K+ tweets/minute with sub-second latency. Collaborated with product teams to operationalize insights into a customer-facing dashboard, increasing user engagement by 15%. Led a team of four engineers to build a real-time technical initiative referenced as “Evolve Squads FlexTrades Technical”.