Headline

Gen AI Developer | AI Engineer | LLMs, RAG, Prompt Engineering & AWS Bedrock | M.Tech Data Science

About

Gen AI Developer with a Master's in Data Science and hands-on experience shipping production LLM systems on AWS. - Most recently, I helped build a Gen AI email-classification platform for an insurance client, where I led prompt engineering and classification accuracy end to end. we drove accuracy from ~40% to 92% across 170+ categories, built the evaluation pipeline and golden dataset behind it, and added PII guardrails (Amazon Bedrock Guardrails) to keep sensitive customer data safe and compliant. - I work across the modern Gen AI stack: LLMs, RAG, prompt engineering and caching, Amazon Bedrock (Claude), and AWS (Lambda, S3, DynamoDB, OpenSearch). I care about building AI that's accurate, fast, cost-efficient, and safe to put in front of real users. Generative AI · Large Language Models (LLMs) · RAG · Prompt Engineering · LLM Guardrails · Amazon Bedrock · AWS · Python · Machine Learning

Location

Greater Bengaluru Area

AI stack

aistudio

anthropic apis

claude code

gemini

claudecode

Supabase

Work experience

Associate Software Engineer at Wise Work, Aug 2024 - Present · 2 yrs 1 mo, On-site, - Built an AI chatbot using LLM function calling to interpret natural-language queries, invoke the right API endpoints, retrieve data dynamically, and return real-time answers to users. - Extended the forecasting system to ingest real-time CPU, memory, and disk utilization data enabling live prediction of resource usage and earlier capacity alerts. - Delivered both as production features within the product, contributing to improved reliability and a smoother user experience.

Gen AI Consultant (Trianz) at Wise Work, Jun 2025 - May 2026 · 1 yr, On-site, Worked as a full-time Gen AI consultant deployed to Trianz for ~1 year, as part of the team that delivered a Gen AI email-classification platform for an insurance client. - Owned prompt engineering for the system, designed, iterated, and optimized the classification prompts that drove accuracy across 170+ categories. - Built an evaluation pipeline on a labeled golden dataset — created with the business team to match the real-world distribution across all 170+ classes — and drove accuracy to 88% on the golden set and 92% in production, tracked over 6 months (up from a legacy system with ~40%). - Implemented prompt caching and automated regex-based classification, helping cut average processing time from 6 to 3 seconds per email with no drop in accuracy, and reducing model inference cost by ~70%. - Implemented guardrails (Amazon Bedrock Guardrails) to detect and mask sensitive customer PII data like names, emails, phone numbers, policy and insurance IDs, before it reached the LLM, supporting compliance for the insurance client. - Contributed to a solution built on AWS (RAG, Lambda, S3, Amazon Bedrock/Claude, DynamoDB, OpenSearch) that reliably processed 1,000+ emails per day. - Recognized with the Team Excellence Award for the project.

Software Engineer Intern at Wise Work, Dec 2023 - Jul 2024 · 8 mos, Bengaluru, Karnataka, India, - Built a time-series forecasting model in Python (ARIMA, SARIMA) to predict system resource utilization, achieving 80%+ accuracy. - Handled the full workflow of data cleaning, feature engineering, and evaluation. Visualized forecasts to support capacity-planning decisions. - Worked with the engineering team to integrate the forecasting output into the product, improving performance and reliability.

Education

Master of Technology - MTech, Data science, kumaraguru college of technology

Bachelor of Engineering - BE, electronics and communication engineering, Sri Eshwar College of Engineering

Public links

github: https://github.com/Sriram-Saikrishnan

leetcode: https://leetcode.com/cssriram10