ANANT SHARMA

AI / GenAI Engineer//

I build production Python systems that think in steps — agentic workflows, retrieval pipelines, and orchestration middleware where every output is gated by automated evals before it ships. 1+ year turning Generative AI research into shipped infrastructure.

View Builds Open Uplink

Review Time

0%cut

Op Cost

0%down

Token Spend

0%saved

Retrieval Recall

0%bench

Monitoring Auto

0%up

deepweave · orchestrator trace streaming

Scroll to descend

[ 01 ]

Core Systems

AI/GenAI engineer shipping production Python systems (FastAPI, Pydantic) across LLMs, Generative AI and agentic AI. I design agentic workflows and workflow-automation solutions with LangChain tool calling, external tool integrations, multi-step reasoning and MCP-aligned APIs that optimize business processes and cut operational cost.

My default posture is measure, then ship. REST API integrations get tested with pytest, LangSmith and Postman; releases go out on Docker + AWS with CI/CD through GitHub Actions; and running systems report into Prometheus and Grafana so regressions surface as alerts, not as user complaints.

What I like most is the seam where a research idea becomes an operational one — taking a promising Generative AI use case through feasibility, demo, stakeholder buy-in, and then into the pipeline that quietly runs every day.

Operator Spec

DesignationAI / GenAI Engineer

Experience1+ Years

Core LangPython · TypeScript

DomainML Engineer specialising in GenAI

RuntimeDocker · AWS · CI/CD

EducationB.Tech CSE (AI/ML)

Status● Operational

[ 02 ]

Field Deployment

AI Software Developer @ CanopusGBS

MAR 2025 → PRESENT

Bangalore, Karnataka · Full-time

−30%Turnaround

Cut enterprise document-review turnaround time across 500+ submission cycles by leading AI/ML solution development of a scalable production RAG pipeline in Python — API integrations (OpenAI API, LangChain, ChromaDB), document chunking, embeddings, vector search and reranking — integrated into existing REST review workflows.

−65%Op Cost

Reduced sector-wise report generation from several hours to under 5 minutes by building a business-process-automation workflow with feature engineering and AI-driven anomaly detection over structured datasets.

+40%Throughput

Increased operational throughput and improved data reliability by developing a Django + SQL backend service with workflow automation, scheduled data-quality automation, cross-team alerting and audit logging — driving continuous process improvement.

3×GenAI Initiatives

Led AI/ML solution development for three GenAI initiatives — research and feasibility analysis for Generative AI use cases, plus demos that aligned solutions with business goals — collaborating with leadership and cross-functional stakeholders across Agile/Scrum sprints to influence technical strategies.

[ 03 ]

Autonomous Builds

UNIT / 001● ACTIVE

DeepWeave

Agentic AI Middleware & Orchestration Layer

Cut token consumption by 35% and LLM API spend by 28% with context-engineering middleware in FastAPI for a LangGraph/LangChain multi-agent orchestration framework — subagent coordination, tool calling and external tool integrations, and asynchronous context handoff across a ReAct-style multi-step reasoning pipeline.
Implemented QA evaluation (guardrail) agents that validate every AI output against business intent, blocking defective responses before they reach production.

PythonFastAPILangGraphLangChain MCP tool-callingAsyncIOMulti-agent

UNIT / 002● ACTIVE

RemembrAI

LLM Memory & Retrieval Agent

Achieved 100% recall across retrieval benchmarks and eliminated ~94% of duplicate record entries by building a semantic vector-search memory agent with conflict-resolution logic.
Gated every release through a CI-style validation layer — automated pytest suites and LangSmith evaluation frameworks for regression gating.

PythonLangGraphVector embeddings Memory mgmtpytestLangSmith

UNIT / 003● LIVE

MLOps Observability Pipeline

Real-time model telemetry & anomaly detection

Increased monitoring automation by 70% and shortened incident detection from manual review cycles to real-time alerts by deploying a live MLOps observability stack.
Anomaly detection and model-performance tracking wired end-to-end across the serving path.

PythonPrometheusGrafana GalileoOpenAI APIAWS

[ 04 ]

Live Lab

Not screenshots. Two real systems running in your browser right now — a neural network training from scratch with hand-written backpropagation, and a working retrieval engine indexing my own résumé. No API calls, no libraries, no server. View source.

neural_forge.js — MLP 2·16·16·1 · tanh · BCE training

decision boundary

Epoch0

Loss—

Accuracy—

Params—

Loss curve

LR 0.45

resume_rag.js — TF·IDF + cosine · 0 chunks indexed

[ 05 ]

Stack Manifest

Generative AI & Agentic

LLMsAgentic workflowsLangChainLangGraph LlamaIndexCrewAIAgents SDKMCP Tool callingMulti-agent orchestrationMulti-step reasoning RAGChunkingEmbeddingsVector searchReranking Prompt engineeringContext managementConversational state & memory Guardrails / output validationNLP · semantic searchPromptfooOpenTelemetry

AI & Machine Learning

Machine LearningAI/ML solution developmentSupervised learning Feature engineeringAnomaly detectionPredictive analytics EDAData processingPyTorchTensorFlowscikit-learn

Backend & Languages

PythonAsyncIO · async/awaitType hintsOOP FastAPIPydanticDjangoREST & MCP-aligned APIs PostmanSQLTypeScriptJavaScriptC / C++

Data & Vector Stores

PostgreSQLMySQLChromaDBPineconeAzure Cosmos DB

LLM Providers

OpenAIAnthropic ClaudeGoogle GeminiDeepSeekLlama

Cloud, DevOps & Evals

AWSDockerCI/CDGitHub ActionsGit pytestLangSmithGalileoPrometheusGrafana

Delivery

Workflow automationBusiness process automationData-quality automation Continuous process improvementAutomation framework design Stakeholder managementSystem designAgile / ScrumTechnical documentation

[ 06 ]

Training Data

2021 — 2025

B.Tech, Computer Science & Engineering (AI/ML Specialisation)

SRM Institute of Science and Technology · Chennai, Tamil Nadu

Certifications

Supervised Machine Learning

DeepLearning.AI & Stanford · 2026

Google AI Essentials

Google · 2026

[ 07 ] · Establish uplink

Let's build
something autonomous.

Open to AI/GenAI engineering roles. If you're shipping agents, retrieval systems or LLM infrastructure into production — I'd like to hear about it.

Emailasindia23@gmail.com Direct+91 88517 79323 LinkedIn/in/codelinechef GitHub@codelinechef