// hi, i'm
◈ Bangalore, India · IST
ML Engineer at SAP Labs building AI agents, RAG pipelines, and intelligent systems. Currently deep in LLMs, GenAI, and MLOps.
01 / about
I'm a Machine Learning Engineer at SAP Labs, Bangalore, working on the GenAI/Joule platform. I build AI agents, RAG pipelines, LangGraph-based agentic systems, and ML-backed tooling for real engineering problems.
With a background spanning full-stack development — Java, Spring Boot, React, Node.js — I bring strong software fundamentals to modern AI systems. Currently pursuing M.Tech in Software Engineering from BITS Pilani.
I've contributed to open source, participated in Devtoberfest, led hackathons, and actively share what I'm learning on LinkedIn. Ask me about Agents, MCP, RAG, or anything GenAI.
02 / now
03 / experience
- Building AI agents using Python, Pydantic, and Joule for enterprise-grade agentic workflows and intelligent automation
- Developing RAG pipelines and agentic systems to solve real engineering problems at scale
- Implementing MLOps practices for model deployment, monitoring, and lifecycle management in production
- Designing agentic workflows with LangGraph for multi-step reasoning and autonomous task orchestration
- Working towards building reusable agent frameworks for scalable business workflow automation
- Built a full-stack stakeholder communication platform with SAP UI5 and Spring Boot across 8–9 capabilities
- Shipped an AI-assisted agent feature using LangChain and Project Agent Builder for context-aware workflows
- Automated PostgreSQL DB migration with Python — auth handling, service keys, tunneling, backup & restore
- Resolved UI and module stability issues, improving reliability and user experience across ISDC
- Led compliance remediation eliminating 65%+ of non-compliant dependencies through upgrades and security patching
- Resolved CI/CD release pipeline failures caused by Git tag-based workflows, restoring stable deployments
- Debugged customer CRMC issues in sandbox environments and traced root causes through application logs
- Created Kibana monitoring alerts for multiple production regions, improving observability across the stack
04 / projects
A Git-inspired memory system for LLM agents with branching, rollback, semantic retrieval, and memory diffing. Features persistent multi-agent memory architecture with vector search, version-controlled state management, and auditable commit history — enabling agents to remember, reason across sessions, and recover from any checkpoint.
End-to-end ML pipeline that automatically classifies support tickets into 22 categories, dramatically cutting manual triage time. Ensemble model (TF-IDF + VotingClassifier with soft voting) achieved 91.75% accuracy and improved real-world predictions from 33% → 100% through smart preprocessing and post-processing. Shipped with a Flask dashboard featuring drag-and-drop batch upload and live Chart.js analytics.
Proactive dashboard monitoring agent with persona-based KPI breach alerts, natural-language explanations, and guided conversational actions. Contextual navigation surfaces relevant views automatically, letting users query insights and trigger workflows without leaving the chat interface.
Adaptive video streaming using HLS enabling automatic bitrate switching based on network conditions. FFmpeg transcoding generates multiple video renditions and HLS playlists. Node.js backend orchestrates video processing, storage, and stream delivery.
05 / skills
06 / writing
07 / education
08 / honors & awards
Let's build something.
Open to collaborations, interesting projects, and conversations about AI, engineering, or anything in between.
rahulkarda2002@gmail.com ↗