Karthik Sherla

Associate Software Engineer and AI/ML Engineer with experience designing and shipping production-grade, agentic GenAI applications, Retrieval-Augmented Generation (RAG) pipelines, and LLM orchestration systems in Python. Hands-on experience with FastAPI, PostgreSQL/MySQL/MongoDB, LangChain, LangGraph, and the Model Context Protocol (MCP) for building intelligent, multi-agent, context-aware chatbots with tool-calling, function calling, and structured outputs. Comfortable across the full ML lifecycle — data preprocessing, model training/fine-tuning, vector and semantic search, and API integration — with deployment practices including Docker, Git-based CI/CD, and AWS.

Track record of measurable impact, including improving model prediction accuracy by up to 94% and reducing data-processing time on production pipelines.

Associate Software Engineer at Codetru (2025-04 – Present)

Progressed from Software Engineering Intern (Apr 2025) to Associate Software Engineer (Sep 2025) on a Payroll platform

• Built and shipped "AI Mode" for the platform — an in-app, tool-calling AI agent that performs actions on a user's behalf and opens a live iframe preview of the application, giving instant visual proof of every change made.
• Designed a production-grade RAG engine with semantic/vector search covering payroll tax rules for all 50 US states, used as a real-time validation layer during payroll runs to catch tax-calculation errors before processing.
• Extended AI and OCR-based document intelligence (powered by the Claude Haiku model) into the Benefits, Deductions, and Reimbursements modules — extracting employee data (salary, dates, bank details) from documents like offer letters and employee sheets, and enabling natural-language, structured-output actions.
• Designed role-based intelligent dashboards (Admin, Owner, Employee) surfacing pending actions — draft/overdue/upcoming payroll and employee/state compliance gaps — plus AI insights panels on payroll, benefits, and reimbursement metrics.
• Implemented priority-action workflows in the Employee module guiding employees and employers through I-9 Section 1 & 2, W-4 submission, and bank verification.
• Explored vision-language and voice/speech models (including Kokoro) to extend the AI assistant toward multimodal use cases.
• Architected a multi-agent LLM chatbot with LLM orchestration to query SQL/NoSQL (MongoDB) in natural language via RAG and the Model Context Protocol (MCP), cutting manual query-writing effort by an estimated 40%.
• Built Python/FastAPI backend services with Redis caching and basic Kafka event pipelines.

Data Science Intern at Codec Technologies (2024-11 – 2025-02)

• Built and deployed end-to-end ML models (Python, Pandas, Scikit-learn) that improved prediction accuracy by up to 94%, enabling data-driven decision-making for stakeholders.
• Designed data pipelines for cleaning, transforming, and analyzing large datasets, reducing overall data-processing time.
• Contributed to an NLP-based project that delivered measurable business impact, including improved efficiency and reduced manual effort.

Python AI/ML Intern at Skill Genie (2024-01 – 2024-05)

• Built and optimized ML models through data preprocessing, feature engineering, training, and evaluation using Python ML libraries.
• Developed RESTful APIs with FastAPI to integrate ML models into scalable production workflows.
• Assisted in fine-tuning Large Language Models (LLMs) and building chatbot solutions focused on performance and accuracy.
• Worked with large datasets, improving data handling, transformation, and visualization to support better decision-making.

Bachelor of Technology (B.Tech) in Computer Science and Engineering – Hyderabad, India (2020-01 – 2023-07)