01
R&D Fresher Intern
Litmus7
Researching memory-aware semantic routing and decentralized graph search for efficient low-latency discovery in dynamic peer networks.
Lokesh YarramalluAI Systems Engineer
01
Research. Build. Ship.
Building dependable AI systems from first-principles research to production delivery.
I work across agentic systems, hybrid retrieval, backend architecture, and production-grade AI workflows with the goal of making advanced systems usable under real constraints.
Current Focus
Multi-agent workflows, MCP-native systems, hybrid RAG, and backend architecture that survives scale and ambiguity.
Education
B.Tech in Computer Science (AI Engineering), Amrita Vishwa Vidyapeetham
9.3
CGPA
Skill Surface
Python (OOP), APIs, SDKs, microservices
Machine Learning, Deep Learning, LLM fine-tuning
LLMOps, AgentOps, MCP, A2A, Hybrid RAG
Docker, Git, FastAPI, Django, WebSockets, GCP
Testing, CI/CD pipelines for GenAI systems
Leadership, communication, problem solving
Operating Pattern
Research
Learn the shape of the problem before forcing a system around it.
Systems
Translate theory into backend, retrieval, orchestration, and deployable workflows.
Delivery
Prioritize resilience, observability, and outcomes over one-off demos.
Experience
Systems experience across research, product, and community leadership.
The work spans research labs, backend deployment, SaaS systems, and AI education, but the through-line is the same: making complex technical ideas operational.
02
Product Development Engineer
Chatpress / HeapVue
Architecting a production-scale multi-tenant AI SaaS with decoupled services, distributed identity, WebSocket communication, and evaluation-driven AgentOps pipelines.
03
Elite Backend Assistant
School of AI, Amrita Vishwa Vidyapeetham
Managed Django deployment on GCP and built a multilingual hybrid RAG system grounded in clinical knowledge for a dermatology AI platform.
04
SIG AI Lead
ACM Student Chapter
Led workshops, national hackathons, and mentoring programs across ML, research, LLMs, and agentic AI.
Featured Work
Proof-of-work shaped like deployable systems.
Four projects that best describe the way I build.
MCP, Playwright, Gemini, AsyncIO
AutoBrowseMCP
Custom MCP server enabling natural-language browser automation with stealth sessions, robust retries, and agent workflow integration.
Agno, Neo4j, Qdrant, Gemini, Cohere
HybridCode
Dual-database code analysis that combines semantic retrieval and structural reasoning for developer questions.
MCP, GRPO Distillation, Hybrid RAG
AMRAssistant
AMR stewardship multi-agent workflow with explanatory reasoning and a custom memory pipeline.
Llama3, PyTorch, XGBoost, oneAPI
HealthAI
Multilingual health assistant with fracture detection and disease prediction optimized for practical deployment.
Publications
BeyondBorders: Unveiling Community and Influence in Multiplex Social Networks
HybridCode: A Dual-Database Framework for Intelligent Codebase Analysis
Deep Learning Enabled Vehicle Identification as a Contribution to Urban Home Security
Industrial Worker Safety Device with Proactive Gas Leak and Fire Protection System
Leadership & Recognition
Finalist, Intel GenAI Hackathon 2024
Winner, MLH Hackathon 2023
Core Organizer, EvoLUMIN National Hackathon
Lead Organizer, EpochOn AI Hackathon
Workshop series on Agentic AI, local LLM deployment, and RAG pipelines designed to bridge theory and production practice.
Closing Note
Open to hard engineering problems where research quality and shipping discipline both matter.
If the work needs agentic systems, backend reliability, hybrid retrieval, or a stronger technical spine, I care about building the version that survives outside the demo.