AGENT-BUILDERACTIVE
LLM agents that finish the job
Agents that go from input to outcome on their own — like the voice AI receptionist I shipped that handles inbound calls end to end, 24/7.
AI AUTOMATION ENGINEER // MUMBAI, IN
Repetitive work is a bug.
Repetitive work is a bug.
I ship the fix_
I build LLM agents, RAG pipelines, and end-to-end automation that ship to production and delete the busywork.
› LLM agents · RAG · n8n workflows · deployed & monitored 24/7
mangesh@prod ~ % systemctl status automation
ai_receptionist.service — ONLINE · uptime 99.9%
content_engine.service — ONLINE · 24/7
lead_pipeline.service — ONLINE
manual_processing.task — TERMINATED · −30% workload
———————————————————————————
✓ 4 systems running · 0 incidents open
mangesh@prod ~ % █
[ PATTERN DETECTED ]
Teams waste hours on work a script could do.
People babysit tasks that should run themselves. I hunt them down, wrap them in LLM-powered automation, and hand the hours back — then keep it alive in production so the work never creeps back.
workflow_autopsy.log
[00:00] INTAKE: manual operations process
[00:01] DETECT: 12 repeated steps / day
[00:03] BUILD: python + LLM automation
[00:05] RESULT: 30% time reclaimed
> STATUS: SHIPPED. RUNNING 24/7.
[ THE ARSENAL ]
Three things I do better than a to-do list.
Three capabilities — one operator who takes them from idea to running software.
PIPELINE-FORGEACTIVE
Automation that removes work
RAG and automation pipelines wiring models, APIs, and data into systems that retire manual work — content engines, lead flows, data jobs.
SHIP & MONITORACTIVE
Built isn't done until it runs
Production deployment on Linux with uptime monitoring and incident response. I sweat the unglamorous part where it has to stay up.
[ BUILD LOG ]
Things I've shipped.
Selected systems — each taken from a blank file to working software.
social_content_engineShipped
Social Content Engine
A full-stack n8n platform: keyword in, platform-ready posts and images out — auto-tuned for Instagram, LinkedIn, X and Facebook.
Pythonn8nOpenAI APIREST APIsPrompt Eng
saleshunterShipped
SalesHunter
Automated lead-gen that fetches and validates prospects via Apify + Apollo, then fires personalised outreach with Selenium + SMTP. Zero manual prospecting.
PythonApifyApollo APISeleniumSMTP
deepfake_detectorIEEE
Deepfake Detector
A Flask backend using a pretrained CNN + cosine similarity to verify faces frame-by-frame. The basis of my IEEE paper.
PythonFlaskOpenCVdlibCNN
letsmeetShipped
LetsMeet
Real-time video conferencing in Django + LiveKit WebRTC — token rooms, live mic/camera controls, and encrypted passphrase handling.
DjangoLiveKitWebRTCJavaScriptPython
[ THE NUMBERS ]
Proof, not adjectives.
0%
Manual time eliminated
across core ops workflows
24/7
Automated coverage shipped
AI receptionist, always on
0
Production systems built
idea → deployed → monitored
0
IEEE publication
peer-reviewed research
[ CORE ENGINE // MODULES LOADED ]
The stack.
ai_agents.moduleLOADED
LLM IntegrationRAGPrompt EngineeringAI Agent WorkflowsVector Databasesn8nAnthropic Claude
languages.moduleLOADED
PythonDjangoFlaskJavaScriptSQLREST APIsHTMLCSSBootstrap
ml_data.moduleLOADED
NumPyPandasScikit-learnOpenCVCNNNLPFeature EngineeringModel EvaluationMatplotlibSeaborn
automation_devops.moduleLOADED
SeleniumApifyWeb ScrapingSMTPMySQLMongoDBGitAWSGCPLinux
[ VERIFIED ]
Published & credentialed.
IEEE RESEARCH PUBLICATION
Deepfake Detection Using Pretrained Model and Cosine Similarity
Peer-reviewed research on verifying faces across video frames with a pretrained CNN and cosine-similarity matching.
IEEE 7th International Conference · 2025
- AWSAWS Academy Graduate — Data EngineeringCloud data pipelines & engineering
- SIMAccenture — Data Analytics & VisualizationJob simulation
- CERT100 Days of Code — Python Pro BootcampAngela Yu · Udemy
[ EDUCATION ]
B.E. Computer Science & Engineering — AI/ML Specialization
Theem College of Engineering · University of Mumbai
Coursework: Artificial Intelligence, Machine Learning, Data Analysis, Software Engineering, Operating Systems, Cloud Computing (AWS).
2021 — 2025
[ SYSTEM QUERIES ]
The honest answers.
What you're probably wondering, minus the fluff.
QUERY_01[ AVAILABILITY ]
Are you open to work?
Yes — open to remote AI & automation roles and freelance builds. Mumbai-based, timezone-flexible. Fastest way in is email.
QUERY_02[ WHAT YOU BUILD ]
What do you actually want to build?
AI agents and automation that ship to production and remove real work. Fewer notebooks, more systems people use every day.
QUERY_03[ HOW YOU WORK ]
How do you work?
End to end. Idea to deployed, monitored, dependable software — and I care most about the part where it has to keep running after launch.
QUERY_04[ STACK ]
Go-to tools?
Python, LLM APIs, n8n, Django/Flask — and Linux when it's time to ship. Whatever gets the system running and keeps it stable.
READY WHEN YOU ARE
Let's ship something.
Open to remote roles, freelance builds, and problems worth automating.
AVAILABLE · MUMBAI, IN · OPEN TO REMOTE