Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

AI/GenAI Engineer with 5+ years of experience building and deploying production-grade LLM systems, agentic AI pipelines, and MLOps/LLMOps infrastructure across financial services at MetLife and JPMorgan Chase. Delivered RAG-powered document intelligence, multi-agent orchestration with LangGraph, and LLMOps monitoring pipelines processing millions of inference requests. Reduced hallucination rates by 34%, cut model deployment cycles from weeks to hours via CI/CD automation, and shipped agentic workflows that eliminated 18,000+ manual analyst hours annually. Hands-on with LangChain, LangGraph, OpenAI, AWS Bedrock, SageMaker, MLflow, Kubernetes, and FastAPI.…AI/GenAI Engineer with 5+ years of experience building and deploying production-grade LLM systems, agentic AI pipelines, and MLOps/LLMOps infrastructure across financial services at MetLife and JPMorgan Chase. Delivered RAG-powered document intelligence, multi-agent orchestration with LangGraph, and LLMOps monitoring pipelines processing millions of inference requests. Reduced hallucination rates by 34%, cut model deployment cycles from weeks to hours via CI/CD automation, and shipped agentic workflows that eliminated 18,000+ manual analyst hours annually. Hands-on with LangChain, LangGraph, OpenAI, AWS Bedrock, SageMaker, MLflow, Kubernetes, and FastAPI.

Neerajakshi Bokka

AI Engineer, Data Engineer, Software Engineer, +2





AI/GenAI Engineer with 5+ years of experience building and deploying production-grade LLM systems, agentic AI pipelines, and MLOps/LLMOps infrastructure across financial services at MetLife and JPMorgan Chase. Delivered RAG-powered document intelligence, multi-agent orchestration with LangGraph, and LLMOps monitoring pipelines processing millions of inference requests. Reduced hallucination rates by 34%, cut model deployment cycles from weeks to hours via CI/CD automation, and shipped agentic workflows that eliminated 18,000+ manual analyst hours annually. Hands-on with LangChain, LangGraph, OpenAI, AWS Bedrock, SageMaker, MLflow, Kubernetes, and FastAPI.…AI/GenAI Engineer with 5+ years of experience building and deploying production-grade LLM systems, agentic AI pipelines, and MLOps/LLMOps infrastructure across financial services at MetLife and JPMorgan Chase. Delivered RAG-powered document intelligence, multi-agent orchestration with LangGraph, and LLMOps monitoring pipelines processing millions of inference requests. Reduced hallucination rates by 34%, cut model deployment cycles from weeks to hours via CI/CD automation, and shipped agentic workflows that eliminated 18,000+ manual analyst hours annually. Hands-on with LangChain, LangGraph, OpenAI, AWS Bedrock, SageMaker, MLflow, Kubernetes, and FastAPI.

Available to hire

AI/GenAI Engineer with 5+ years of experience building and deploying production-grade LLM systems, agentic AI pipelines, and MLOps/LLMOps infrastructure across financial services at MetLife and JPMorgan Chase. Delivered RAG-powered document intelligence, multi-agent orchestration with LangGraph, and LLMOps monitoring pipelines processing millions of inference requests.

Reduced hallucination rates by 34%, cut model deployment cycles from weeks to hours via CI/CD automation, and shipped agentic workflows that eliminated 18,000+ manual analyst hours annually. Hands-on with LangChain, LangGraph, OpenAI, AWS Bedrock, SageMaker, MLflow, Kubernetes, and FastAPI.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Language

Work Experience

AI/GenAI Engineer at MetLife

August 1, 2024 - Present

Architected a multi-agent LangGraph pipeline for policy document Q&A and claims triage, integrating GPT-4o and AWS Bedrock Claude, reducing analyst review time by 46% and processing 2M+ policy pages per month. Built and deployed a RAG system using FAISS and LangChain over underwriting and actuarial knowledge bases, achieving 34% lower hallucination rate (RAGAS-evaluated) and cutting manual research time by 3.2 hours per underwriter per week. Established an end-to-end LLMOps monitoring stack with MLflow, Prometheus, and Grafana, tracking token drift, latency SLAs, and prompt regression across 12 production models, enabling zero-downtime rollbacks on model version updates. Engineered prompt hardening and PII redaction using SpaCy NER and regex classifiers for SOX/PII compliance before inference, blocking 1,400+ sensitive data leakage events. Containerized AI microservices with Docker and Kubernetes (EKS) using FastAPI endpoints, reducing model deployment cycle from 3 weeks to under 6 hou

AI Engineer Intern (GenAI Applications, MLOps, Fraud Intelligence) at JPMorgan Chase

January 1, 2024 - July 31, 2024

Delivered a fine-tuned LoRA adapter on LLaMA-3 for transaction narrative classification across payment and retail banking systems, improving fraud signal precision to 91.3% F1 across 27 fraud typologies, supporting prevention of 14 confirmed fraud attempts without customer fund loss. Built a multimodal anomaly detection pipeline combining tabular ML (XGBoost) with LLM-generated explanations via GPT-4 Turbo, reducing Tier-1 analyst escalations by 41% and cutting mean time to investigate from 22 to 9 minutes. Deployed MLflow experiment tracking and model registry across 3 AI squads, standardizing 6,500+ daily inference runs on SageMaker with auto-scaling and A/B shadow deployment, cutting staging-to-production drift incidents by 38%. Authored an internal LLM evaluation framework using LangSmith and custom benchmark suites covering factuality, toxicity, and task adherence, validating 9 model rollouts before production and reducing post-release regressions by 30%.

Software Engineer, Backend Python at Adons Softech

March 1, 2019 - July 31, 2023

Designed and owned RESTful microservices in Python/FastAPI for a multi-tenant SaaS platform serving 12+ enterprise clients, handling 400K+ daily API requests with p99 latency under 120ms. Re-architected a legacy monolith into event-driven services using Kafka and Celery, reducing background job failure rate from 8.4% to under 0.3% and enabling horizontal scaling during peak ingestion windows. Built data pipeline modules in Python (Pandas, SQLAlchemy) for ETL workflows across PostgreSQL and Elasticsearch, cutting nightly report generation time from 4.5 hours to under 35 minutes. Implemented JWT/OAuth2 auth layer and RBAC enforcement across all internal APIs, eliminating 100% of unauthorized cross-tenant data access incidents during quarterly security reviews for 12+ client environments. Wrote unit and integration test suites (pytest, 87%+ coverage) automated via GitHub Actions CI, reducing regression bugs in production by 35% across 6 product releases.

Software Engineer (Backend Python) at Adons Softech

March 1, 2019 - July 31, 2023

Designed and owned RESTful microservices in Python/FastAPI for a multi-tenant SaaS platform serving 12+ enterprise clients, handling 400K+ daily API requests with p99 latency under 120ms. Re-architected a legacy monolith into event-driven services using Kafka and Celery, reducing background job failure rate from 8.4% to under 0.3% and enabling horizontal scaling during peak ingestion windows. Built ETL data pipeline modules in Python (Pandas, SQLAlchemy) across PostgreSQL and Elasticsearch, cutting nightly report generation time from 4.5 hours to under 35 minutes. Implemented JWT/OAuth2 authentication and RBAC enforcement across internal APIs, eliminating unauthorized cross-tenant data access incidents during quarterly security reviews for 12+ client environments. Wrote unit and integration tests (pytest, 87%+ coverage) automated via GitHub Actions CI, reducing regression bugs in production by 35% across 6 product releases.