available for workAI Engineer — GenAI & Full-Stack
ChiragSingh.
I build and operate LLM systems in production — retrieval, evals, and serving that hold up under real load. Less demo, more infrastructure.
1.8 yrs
engineering AI
12
shipped projects
1
publication
~/chirag — zsh
$ type whoami anywhere ↵
// live systems
synced just now
github.com/chirag405
open ↗0
contributions · 1y
0
public repos
0
stars earned
0
day streak
top languages
recent commits
leetcode/chirag406
open ↗svc/genai-support-bot
healthy0
tokens processed · 24h
api calls · 24h
$0.0000
cost / query
0ms
p95 latency
throughput · tok/s
no traffic yet — say hi to ask_chirag ↓
// GitHub + LeetCode are fetched live; the observability card tracks this site's own ask_chirag assistant.
// experience
2 roles
July 2025 — Present
Associate GenAI & Full-Stack Engineer@ IncerroArchitected the production RAG pipeline end to end — retrieval, reranking, and evaluation gating — extending it with GraphRAG and shipping agentic workflows, full-stack surfaces, and Meta Quest XR rendering along the way.
what I did
→Architected the production RAG pipeline end to end — retrieval, reranking, evaluation gating — and extended it with GraphRAG for relationship-aware retrieval, improving answer relevance by 20% and multi-hop query accuracy by 25%
→Built agentic AI workflows in LangGraph instrumented with Langfuse tracing and evals, cutting token consumption by 30% and p95 response latency by 35% while raising output quality scores
→Developed MCP (Model Context Protocol) servers to standardize internal tool and data access for LLM agents, cutting new tool onboarding time by 40%
→Shipped production full-stack surfaces in Next.js, FastAPI, and React Native — including internal prompt-versioning and evaluation tooling — reducing prompt iteration cycle time by 50%
→Engineered XR rendering for Meta Quest using Meta Spatial SDK and Vulkan on native Android, sustaining 72 FPS under target frame-time budgets
PythonLangChainLangGraphLangfuseGraphRAGMCPvLLMFastAPINext.jsReact NativePostgresMeta Spatial SDKVulkan
June 2024 — Sept 2024
Flutter Developer Intern@ Higroove SystemsHybrid internship — cross-platform mobile development in Flutter/Dart, shipping on-device AI features with TensorFlow Lite and OpenCV across Android and iOS.
what I did
→Developed and maintained cross-platform mobile applications in Flutter/Dart, delivering consistent performance and UI parity across Android and iOS
→Built and integrated on-device AI models using TensorFlow Lite and OpenCV for real-time image-based features, enabling inference without network dependency
→Optimized application performance through profiling and bug remediation, reducing cold-start load time by 32% and crash rate by 24%
FlutterDartTensorFlow LiteOpenCVREST APIsAndroidiOS
// selected work
12 shipped projects
// publications
[1]
Plate Vision: A Number Plate Recognition Using AI/ML and YOLOv8
C. Singh et al. et al. · International Journal of Innovative Research in Computer and Communication Engineering (IJIRCCE) · Vol. 13, Issue 5 · May 2025.
// stack
Languages
TypeScriptJavaScriptPythonJavaDartSQLKotlinC/C++
Generative AI / LLM
LangChainLangGraphLlamaIndexFine-tuningRAGAI AgentsAgentic WorkflowsOrchestrationPrompt EngineeringContext EngineeringMemory ManagementLLM-as-a-JudgeEmbeddingsVector SearchEvalsvLLM
Machine Learning / Computer Vision
TensorFlowScikit-learnYOLOv8OpenCVTesseractHugging FaceComputer VisionObject DetectionImage ClassificationOCRImage ProcessingClassical MLModel TrainingModel EvaluationInference
Backend
Node.jsExpress.jsFastAPISpring BoottRPCPostgreSQLMySQLRedis
Frontend
ReactNext.jsTailwind CSSZustandRedux
Mobile
FlutterReact NativeExpo
XR / 3D / Spatial Computing
Three.jsReact Three FiberWebXROpenXRAndroid XRAndroidX XRSceneCoreMeta Spatial SDKNiantic Spatial SDKBabylon.jsA-FrameWebGL
Cloud & Infrastructure
DockerKubernetesAWSGCPCloudflare WorkersVercelNetlifyAWS AmplifyCI/CDLangfusePhoenix
Developer Tools
GitGitHubClerkInngestBubblewrapChrome DevToolsAndroid NDKCMakeSpector.jsWebGL Inspector
Testing
VitestPlaywright