Staff Engineer for applied AI and backend systems.

I lead engineering for production AI systems and build platforms that handle 100M+ annual label uploads.

Two AI systems in daily client use in ~7 weeks
100M+ annual label uploads
~190K daily LLM completions
100K+ app downloads
20K+ monthly active users
30% lower compute costs
5× deployment velocity
8 engineers hired and mentored
Two AI systems in daily client use in ~7 weeks
100M+ annual label uploads
~190K daily LLM completions
100K+ app downloads
20K+ monthly active users
30% lower compute costs
5× deployment velocity
8 engineers hired and mentored
Two AI systems in daily client use in ~7 weeks
100M+ annual label uploads
~190K daily LLM completions
100K+ app downloads
20K+ monthly active users
30% lower compute costs
5× deployment velocity
8 engineers hired and mentored

Hackathon Wins

Best Use of Amazon Bedrock

Mentation

MongoDB AI Hackathon (5,000+ participants)

Product Design: Mrinmoyee Banerjee

AI-powered mental health companion with journaling, emotion tracking, and personalized support. Built to make therapy accessible in India.

AWS BedrockSageMakerNext.jsMongoDBCognito
Pathsure
Best PHI & PII Solution

Pathsure

OpenText Hacknosis Hackathon (Top 3% of 1968 teams)

Product Design: Mrinmoyee Banerjee

Secure medical document management system. Enables patients to control medical history sharing and helps doctors extract relevant info for faster diagnoses.

OpenText APIsAndroidWeb AppSecurity

Selected Projects

Move Onn
700 customers

Move Onn

Side Project

Launched an AI breakup-recovery app for iPhone and iPad with 700 customers and $200 in monthly recurring revenue.

iOSAI chatVoice
250+ users

Trott

Side Project

Product Design: Mrinmoyee Banerjee

Built semantic search and retrieval-augmented chat for a short-video organizer. Find saved reels and shorts by the places, recipes, and products they mention. Built for RevenueCat Shipathon 2025.

FlutterFastAPIGCPLLMs
Capywarz
Solo Dev

Capywarz

Game Dev

Multiplayer game built solo in Unity in 30 days. Complete game loop with mechanics and progression.

UnityC#Game Networking
Distributed Label Pipeline
10M images in 30 min

Distributed Label Pipeline

Checko

Built a distributed pipeline for product authentication labels with computer vision, two-stage human review, and cloud sync. Cut processing time for 10 million images from 1–4 days to 30 minutes.

PythonFastAPIGCP Cloud RunPostgreSQLCV
Private Repo
Checko App
100K+ Downloads

Checko App

Checko

Scaled consumer apps to 100K+ downloads and 20K+ monthly active users across four countries.

KotlinAndroidFirebaseOptimization

Experience

Methodic Labs (Deployly)

Sep 2026 – Present
Staff Engineer
  • Led architecture and delivery for two AI systems that reached daily client use in about seven weeks.
  • Built a Twilio/LiveKit agent that confirms appointments for two pediatric therapy clinics and writes results back to Raintree. Added a deploy gate that blocks releases during live calls and 450+ backend tests.
  • Built a Next.js, PostgreSQL, and Claude platform that turns research briefs into questionnaires and Decipher survey XML, with deterministic logic checks and an $8 model budget per study.
  • Built Anderson’s Salesforce Lightning intake for land trusts and deeds with sandbox-verified prefill. Backtested an AI validation prototype on 300 historical cases.

Checko (Transpacks Technologies Pvt Ltd)

Apr 2021 – Present
Founding EngineerHyderabad, India
  • Built a product authentication platform handling 100M+ annual label uploads. Cut processing time for 10 million images from 1–4 days to 30 minutes with a distributed pipeline.
  • Scaled consumer apps to 100K+ downloads and 20K+ monthly active users across four countries.
  • Cut compute costs by 30% and halved technology spend through autoscaling and usage-based policies. Helped raise gross margin by 5 percentage points.
  • Hired and mentored eight engineers. Increased deployment velocity 5× with trunk-based CI/CD.

Creed

Nov 2025 – Aug 2026
Staff Engineer
  • Architected a multi-provider LLM layer for chat, voice, and memory. Production logs show ~190K daily completions and ~6K recovered by automatic failover.
  • Rebuilt a Celery and Vertex Gemini pipeline handling ~700K daily pushes. Streamed inputs to prevent memory failures and gated stale chats to cut ~130K daily model requests without reducing sends.
  • Improved mem0 and pgvector retrieval to surface 24% more memories in a 30-user sample. Cut worst-case conversation compaction from 44s to 0.5s across 113 million messages.

NCFlexE, IIT Kanpur

Jun 2020 – Mar 2021
Project Associate and Summer InternKanpur, India
  • Delivered 4+ R&D projects across ML, automation, and backend.
  • Built Android apps with OpenCV, TensorFlow, and ML Kit.
  • Managed GCP/AWS services for datasets with 100K+ records.

Boston Scientific

Jan 2020 – Jun 2020
Software Development Engineer InternGurugram, India
  • Developed internal web apps using PrimeFaces and ThingWorx.
  • Integrated Windchill data via Swagger REST APIs.

The Sparks Foundation

May 2019 – Aug 2019
Project Intern

Manipal Institute of Technology

May 2019 – Aug 2019
Summer Intern

Startup Consulting

Ongoing
Technical ConsultantUS / Singapore / India
  • Consulting for startups ranging from pre-seed to mid-sized stages.
  • Advising on technical architecture, scalability, and engineering practices for distributed teams.
  • Helping founders navigate 0-to-1 engineering challenges and MVP launches.

Tech Stack

Languages

PythonKotlinTypeScript

Backend

FastAPINext.jsCelerygRPCPostgreSQLpgvectorMongoDB

AI and Voice

ClaudeVertex AILiveKitTwilio

Cloud

GCP Cloud RunAWSKnativeDocker

DevOps

GitHub ActionsPrometheusGrafana

Mobile

React NativeAndroid SDKFlutterFirebase

About

At Methodic Labs, I led architecture and delivery for a voice agent and a survey drafting platform that both reached daily client use in about seven weeks. At Creed, I built backend systems for AI chat, voice, memory, and personalized notifications.

As a founding engineer at Checko, I built a product authentication platform that handles 100M+ annual label uploads. I hired and mentored eight engineers, and our work on autoscaling and usage-based policies cut compute costs by 30%.

I earned a B.Tech in Computer Science from Sikkim Manipal Institute of Technology in 2020 (8.13/10). I also served as president of the Photography and Art Clubs.

Interests

TravelingOil PaintingsPotteryHikingProcess AutomationGame Dev

Sudipta Dey

Hyderabad, IN

Let's build something
together.