Skip to content
Here's what I actually ship.

Lohith Kovuri

Software Engineer working across Backend, Distributed Systems & AI/LLM Platforms

Backend • Distributed Systems • Cloud (AWS) • AI/LLM Platforms

PythonJavaAWSFastAPILangChain/LangGraph

I build fault-tolerant backend systems and production AI/LLM platforms, including event-driven AWS pipelines that process tens of thousands of events a day, and multi-tenant RAG systems that serve hundreds of concurrent tenants without leaking a byte between them.

View ResumeRecruiter ModeExplore Projects
GitHubLinkedInContact Me
write availability
0.0%

write availability

events/day processed
0+

events/day processed

concurrent RAG tenants
0+

concurrent RAG tenants

fewer redundant LLM calls
0%

fewer redundant LLM calls

Why me

Evidence, not adjectives.

Five things I actually shipped, with the numbers that prove it.

99.9%
write availability

Operated production AWS infrastructure at Mercury (a fintech platform) sustaining 99.9% write availability across 50,000+ events/day, including 8x traffic bursts, with zero data loss.

200+
concurrent tenants

Architected a multi-tenant RAG backend serving 200+ concurrent tenants with per-tenant vector-store isolation (zero cross-tenant data leaks in stress testing), while cutting redundant LLM calls by 62% and dropping latency from 1.8s to 680ms.

120s → <60s
p95 latency

Took a synchronous LLM pipeline from p95 latency of 120s down to under 60s by rebuilding it as an async, backpressure-aware FastAPI service.

90%+
clause-extraction accuracy

Shipped an agentic document-intelligence pipeline (VendorOps AI, LangGraph, public repo) achieving 90%+ clause-extraction accuracy on the 100+ contract CUAD legal benchmark, at 60% lower LLM cost per document via small/large-model routing.

3
cloud & data certifications

Credentialed on the cloud and data stack he ships on: AWS Certified Data Engineer – Associate, AWS Certified Cloud Practitioner, and Microsoft Certified: Power BI Data Analyst Associate.

Project showcase

Engineering case studies

7 systems I designed and built. Tap any card for the full problem → architecture → impact breakdown.

Problem

Manual code review doesn't scale with PR volume, and most "AI code review" tools are single-shot LLM calls with no way to recover if one step fails partway through.

Solution

A multi-tool agent system (diff retriever, static analyzer, test generator, and PR commenter) with tool-call retry and state checkpointing, so a mid-chain failure resumes from the last completed tool instead of restarting the whole review.

Engineering challenge

Keeping the orchestration loop model-agnostic while making a 4-tool agent chain resilient to mid-run failures, via state checkpointing rather than an all-or-nothing retry.

Architecture

  1. GitHub Actions (on pull_request)
  2. Agent orchestrator (LangGraph, model-agnostic)
  3. Diff retriever
  4. Static analyzer
  5. Test generator
  6. PR commenter
  7. GitHub PR comment API

Impact

Catches 85%+ of seeded defects in benchmark diffs, at under $0.05 per review, running automatically on every pull request.

Tech

PythonLangGraphFastAPIOpenAI APIGitHub Actions

Closed source. No public repository for this project.

Also on GitHub

Actively-maintained side projects

View all repositories on GitHub
meeting-intelligence-agent

In-progress AI agent for meetings, split into service-oriented modules (transcription, extraction, agent orchestration, memory).

Shows current hands-on work with agent architecture and service decomposition, actively updated.

Python
Linkedin-post-generator

LangChain + Groq-hosted LLaMA app that generates LinkedIn posts (English/Telugu) from topic tags using few-shot prompting, served via Streamlit.

Shows practical LLM prompting/product work end-to-end, shipped and usable.

Python

Toolbox

What I can do with it

Organized by engineering capability, not a wall of logos. No arbitrary percentage bars, just what I actually use and what it lets me build.

Build

Languages

PythonJavaTypeScriptJavaScriptRustC++SQL

Architect

Backend & APIs

FastAPISpring BootFlaskNode.js/ExpressREST APIsAsync & event-driven microservices

Reason

AI / GenAI

LangChainLangGraphOpenAI APIRAG pipelinesFAISSMCP (Model Context Protocol)RAGAS evaluationPrompt engineeringMulti-agent orchestration

Ship

Cloud & DevOps

AWS (ECS, Lambda, SQS/SNS, EC2, RDS, S3, CloudWatch)Azure (Service Bus, Functions, CosmosDB)DockerKubernetesTerraformGitHub ActionsJenkinsCI/CD

Flow

Data & Streaming

KafkaFlinkSparkAirflowPostgreSQLDynamoDBRedisMySQL

Interface

Frontend

ReactTypeScript

Verify

Practices

Agile/ScrumCode reviewDistributed tracingObservabilityPytestPostmanJest

Experience

Where I've shipped

    • Designed a fault-tolerant, distributed event-ingestion service on AWS with idempotent writes and dead-letter replay, sustaining 99.9% write availability across 8x traffic bursts.
    • Reduced MTTR by 40% by redesigning failure-recovery logic with exponential backoff and documenting recovery procedures per failure mode.
    • Built an AI-powered test-generation system in CI that synthesizes edge cases from schema definitions across 3 services, cutting development time by 60%.
    • Instrumented CloudWatch multi-signal observability and anomaly-detection alarms, cutting mean incident detection time from hours to minutes and shifting the team from reactive to proactive response.
    AWS (ECS, S3, SQS/SNS, Lambda, CloudWatch)OpenAI APICI/CD

Background

Education & certifications

Education

Rowan University

Sep 2023 – May 2025

M.S. Data Science

Concentration: Statistical ML, Data Engineering

Glassboro, NJ

Dr. MGR Educational and Research Institute

Sep 2019 – May 2023

B.Tech Information Technology

Chennai, India

Certifications

AWS Certified Data Engineer – Associate

Amazon Web Services

AWS Certified Cloud Practitioner

Amazon Web Services

Microsoft Certified: Power BI Data Analyst Associate

Microsoft

Get in touch

Let's talk about your engineering team.

San Francisco, CA, USA, Open to new opportunities

Emaillohithkovuri321@gmail.comLinkedInlinkedin.com/in/contact-lohithkovuriGitHubgithub.com/LOHITHK22ResumeLohith_Kovuri_Resume.pdf
Download Resume