Open to interesting problems

Building AI systems
that ship at scale.

AI/ML Engineer crafting production computer-vision pipelines and infrastructure for AI visual merchandising software serving automotive marketplaces across the US, Canada & India.

View Resume See My Work
✉ GH in K
3+
Years in Production AI
99%
API Latency Reduction
15+
CV Models Fused in Prod
$64K+
Annual Cost Savings Driven
Career

Professional Experience

From individual contributor to technical lead — owning the applied-AI lifecycle end to end.

AI/ML Engineer
HelloRamp [HelloAR] — Bengaluru, India · AI visual merchandising software
July 2023 — Present

Computer Vision & Applied AI

  • Built an end-to-end virtual-studio car imaging pipeline fusing 15+ production CV models to output photorealistic images at 97–99.5% accuracy/IoU.
  • Scaled to ~250K+ images/month with 8x burst capacity, cutting per-image processing time 70% and memory footprint 95%+.
  • Shipped a thin-part detection model for buses, raising accuracy from 60% to 85% while cutting inference latency ~4x and container size ~15x.
  • Reduced model training compute cost 75% and distributed training time from ~3 months to under a week.
  • Automated 3D synthetic-data generation (Blender, multi-GPU), cutting a 450+ day render workload to a week.
  • Engineered an interactive 360° car spin pipeline using novel-view-synthesis and multi-view geometry.
  • Evaluated and prototyped generative-model approaches for background inpainting, reflection removal, and image harmonization.

Backend, Cloud & Platform

  • Drove a company-wide cost-reduction program delivering over $64,000/year in recurring savings.
  • Cut GPU inference hosting cost ~7x via a self-hosted multi-model inference platform across two cloud migrations.
  • Led an auth performance overhaul cutting response time 99%+ to under 100ms.
  • Implemented a GCP AI inference gateway across 30+ endpoints with Redis rate-limiting and metrics-driven autoscaling.
  • Established MongoDB schemas, aggregation pipelines, and async task-queue infrastructure across microservices.
  • Owned production incident response and database reliability at scale, including a retention cleanup that cut one major collection's size by a third.

Team Leadership & Delivery

  • Advanced from individual contributor to a technical lead role; mentored four AI and backend engineers while still shipping individual technical work end-to-end.
  • Owned the AI/CV technical roadmap and technical hiring; handled technical support escalations for major enterprise accounts.
  • Guided delivery of an automated social pipeline (AI avatar/script/voice generation) and internal analytics dashboards.
  • Steering an automated image quality-check and auto-correction system with AI-flagged human review routing.
Maintenance Engineer, Graduate Engineer Trainee
Reliance Industries Limited — Hazira, Gujarat · Mechanical plant maintenance at a large-scale manufacturing facility
2019 — 2020
Currently Building

Projects

🤖
In Progress

AI-Powered Job Application Agent

Aug 2026 — Present

Agentic system automating the end-to-end job application workflow: searches jobs, parses descriptions, matches candidate profile via RAG, generates tailored materials, and completes submissions through browser automation.

LangGraphRAGMCPFastAPIPlaywright
More Projects →
Credentials

Certifications

Key Coursera specializations most relevant to my current work.

Applied Data Science Specialization↗ Deep Learning Specialization↗ DeepLearning.AI TensorFlow Developer Specialization↗ IBM Applied AI Specialization↗ IBM Data Science Specialization↗ Reinforcement Learning Specialization↗
View All Certifications →
Background

About & Education

I started out as a Mechanical Engineer at VNIT Nagpur, then spent a year in plant maintenance at Reliance Industries before pivoting into programming and data science — completing a second Bachelor's degree at IIT Madras while building a career in applied AI from the ground up.

Since 2023 I've been at HelloRamp [HelloAR], where I progressed from individual contributor to technical lead, building the computer-vision systems, backend platform, and cloud infrastructure behind an AI visual merchandising product used by automotive marketplaces across three countries.

I care about production reliability as much as model accuracy — most of the work above is as much about cost, latency, and uptime as it is about pushing pixels through a neural network.

B.Sc., Programming and Data Science
Indian Institute of Technology (IIT), Madras · CGPA 9.2/10
2020 — 2023
B.Tech, Mechanical Engineering
Visvesvaraya National Institute of Technology (VNIT), Nagpur · CGPA 8.12/10
2015 — 2019

📄 Resume

View the full HTML version in-browser, or download the PDF — always up to date with my latest role and impact.

View Resume

Let's talk

Open to interesting problems in AI, computer vision, and backend systems.

tushar@portfolio: ~
tushar@dev:~$ whoami

> Tushar Supe

role: AI/ML Engineer @ HelloRamp [HelloAR]

Building production computer-vision pipelines and backend infra for AI visual merchandising software serving automotive marketplaces across the US, Canada & India. From individual contributor to technical lead.

PythonPyTorchComputer Vision MongoDBAWS / GCPLangGraph
▶ run resume.html cat projects/
3+ yrsproduction AI
99%latency cut
15+CV models fused
$64K+saved / year
tushar@dev:~$ cat experience.log
AI/ML Engineer
Jul 2023 — Present
HelloRamp [HelloAR] — Bengaluru, India · AI visual merchandising software
computer_vision_and_applied_ai
  • Built an end-to-end virtual-studio car imaging pipeline fusing 15+ production CV models at 97–99.5% accuracy/IoU.
  • Scaled to ~250K+ images/month, 8x burst capacity; cut per-image processing time 70%, memory footprint 95%+.
  • Shipped bus thin-part detection: accuracy 60%→85%, inference latency ~4x faster, container size ~15x smaller.
  • Reduced model training compute cost 75%; distributed training time from ~3 months to under a week.
  • Automated 3D synthetic-data generation (Blender, multi-GPU): 450+ day render workload → a week.
  • Engineered an interactive 360° car spin pipeline via novel-view-synthesis and multi-view geometry.
  • Prototyped generative-model approaches for background inpainting, reflection removal, and image harmonization.
backend_cloud_and_platform
  • Drove a company-wide cost-reduction program: $64,000+/year in recurring savings.
  • Cut GPU inference hosting cost ~7x via self-hosted multi-model platform, two cloud migrations.
  • Led an auth overhaul cutting response time 99%+ to under 100ms.
  • Implemented a GCP AI inference gateway across 30+ endpoints with Redis rate-limiting + autoscaling.
  • Established MongoDB schemas + async task-queue infra across microservices.
  • Owned production incident response and DB reliability at scale, including a retention cleanup that cut one major collection's size by a third.
team_leadership_and_delivery
  • Advanced from individual contributor to a technical lead role; mentored four AI and backend engineers while still shipping individual technical work end-to-end.
  • Owned the AI/CV roadmap and technical hiring; handled technical support escalations for major enterprise accounts.
  • Guided delivery of an automated social pipeline (AI avatar/script/voice) + internal analytics dashboards.
  • Steering an automated image quality-check + auto-correction system with human-review routing.
Maintenance Engineer, Graduate Engineer Trainee
Reliance Industries Limited — Hazira, Gujarat · mechanical plant maintenance
2019 — 2020
tushar@dev:~/projects$ ls -la --sort=date | head -1
🤖
in_progress
ai_job_application_agent
Aug 2026 — Present

Agentic system automating the end-to-end job application workflow: searches jobs, parses descriptions, matches candidate profile via RAG, generates tailored materials, and completes submissions through browser automation.

LangGraphRAGMCPFastAPIPlaywright
▶ ls ~/projects/all
tushar@dev:~$ cat certifications.log
Applied Data Science Specialization↗ Deep Learning Specialization↗ DeepLearning.AI TensorFlow Developer Specialization↗ IBM Applied AI Specialization↗ IBM Data Science Specialization↗ Reinforcement Learning Specialization↗
▶ view all certifications
tushar@dev:~$ cat about.md

I started out as a Mechanical Engineer at VNIT Nagpur, then spent a year in plant maintenance at Reliance Industries before pivoting into programming and data science — completing a second Bachelor's degree at IIT Madras while building a career in applied AI from the ground up.

Since 2023 I've been at HelloRamp [HelloAR], where I progressed from individual contributor to technical lead, building the computer-vision systems, backend platform, and cloud infrastructure behind an AI visual merchandising product used by automotive marketplaces across three countries.

I care about production reliability as much as model accuracy — most of the work above is as much about cost, latency, and uptime as it is about pushing pixels through a neural network.

B.Sc., Programming and Data Science
IIT Madras · CGPA 9.2/10
2020 — 2023
B.Tech, Mechanical Engineering
VNIT Nagpur · CGPA 8.12/10
2015 — 2019

open resume.html

View the full HTML version in-browser, or download the PDF — always up to date.

▶ view resume

contact --info

Open to interesting problems in AI, computer vision, and backend systems.

🚀 8x burst capacity
🧠 15+ CV models fused
⚡ <100ms API latency
👋 Available for new opportunities

Hi, I'm Tushar —
AI/ML Engineer

I build production computer-vision pipelines and scalable backend systems for AI visual merchandising software at an automotive-focused startup.

📄 View Resume 💼 See My Work
3+
Years in Production AI
99%
API Latency Cut
15+
CV Models Fused
$64K+
Saved / Year
Career

Professional Experience

From individual contributor to technical lead — owning the applied-AI lifecycle end to end.

AI/ML Engineer
HelloRamp [HelloAR] — Bengaluru, India · AI visual merchandising software
July 2023 — Present

Computer Vision & Applied AI

  • Built an end-to-end virtual-studio car imaging pipeline fusing 15+ production CV models at 97–99.5% accuracy/IoU.
  • Scaled to ~250K+ images/month with 8x burst capacity, cutting processing time 70% and memory footprint 95%+.
  • Shipped bus thin-part detection: accuracy 60%→85%, inference ~4x faster, container size ~15x smaller.
  • Reduced model training compute cost 75%; training time from ~3 months to under a week.
  • Automated 3D synthetic-data generation (Blender, multi-GPU): 450+ day render job → a week.
  • Engineered a 360° car spin pipeline via novel-view-synthesis and multi-view geometry.
  • Prototyped generative-model approaches for background inpainting, reflection removal, and image harmonization.

Backend, Cloud & Platform

  • Drove a company-wide cost-reduction program: $64,000+/year in recurring savings.
  • Cut GPU inference hosting cost ~7x via self-hosted multi-model platform.
  • Led an auth overhaul cutting response time 99%+ to under 100ms.
  • Implemented a GCP AI inference gateway across 30+ endpoints with autoscaling.
  • Established MongoDB schemas and async task-queue infra across microservices.
  • Owned production incident response and DB reliability at scale, including a retention cleanup that cut one major collection's size by a third.

Team Leadership & Delivery

  • Advanced from individual contributor to a technical lead role; mentored four AI and backend engineers while still shipping individual technical work end-to-end.
  • Owned the AI/CV roadmap and technical hiring; handled technical support escalations for major enterprise accounts.
  • Guided delivery of an automated social pipeline and internal analytics dashboards.
  • Steering an automated image quality-check + auto-correction system with human review routing.
Maintenance Engineer, Graduate Engineer Trainee
Reliance Industries Limited — Hazira, Gujarat · mechanical plant maintenance
2019 — 2020
Currently Building

Projects

🤖
In Progress

AI-Powered Job Application Agent

Aug 2026 — Present

Agentic system automating the end-to-end job application workflow: searches jobs, parses descriptions, matches candidate profile via RAG, generates tailored materials, and completes submissions through browser automation.

LangGraphRAGMCPFastAPIPlaywright
More Projects →
Credentials

Certifications

Key Coursera specializations most relevant to my current work.

Applied Data Science Specialization↗ Deep Learning Specialization↗ DeepLearning.AI TensorFlow Developer Specialization↗ IBM Applied AI Specialization↗ IBM Data Science Specialization↗ Reinforcement Learning Specialization↗
View All Certifications →
Background

About & Education

I started out as a Mechanical Engineer at VNIT Nagpur, then spent a year in plant maintenance at Reliance Industries before pivoting into programming and data science — completing a second Bachelor's degree at IIT Madras while building a career in applied AI from the ground up.

Since 2023 I've been at HelloRamp [HelloAR], where I progressed from individual contributor to technical lead, building the computer-vision systems, backend platform, and cloud infrastructure behind an AI visual merchandising product used by automotive marketplaces across three countries.

I care about production reliability as much as model accuracy — most of the work above is as much about cost, latency, and uptime as it is about pushing pixels through a neural network.

B.Sc., Programming and Data Science
Indian Institute of Technology (IIT), Madras · CGPA 9.2/10
2020 — 2023
B.Tech, Mechanical Engineering
Visvesvaraya National Institute of Technology (VNIT), Nagpur · CGPA 8.12/10
2015 — 2019

📄 Resume

View the full HTML version in-browser, or download the PDF — always up to date with my latest role and impact.

View Resume

Let's talk

Open to interesting problems in AI, computer vision, and backend systems.

Available for new opportunities

Shipping AI
at production scale.

AI/ML Engineer bridging computer vision research and production infrastructure. Specializing in applied computer vision and backend systems for AI-powered imaging software.

View Resume → GitHub
3+
Years Exp.
15+
CV Models Shipped
99%
API Latency Cut
$64K+
Saved / Year
// career

Experience Log

From individual contributor to technical lead — owning the applied-AI lifecycle end to end.

July 2023 — Present
AI/ML Engineer
HelloRamp [HelloAR] — Bengaluru, India · AI visual merchandising software

Computer Vision & Applied AI

  • Built an end-to-end virtual-studio car imaging pipeline fusing 15+ production CV models at 97–99.5% accuracy/IoU.
  • Scaled to ~250K+ images/month with 8x burst capacity, cutting processing time 70% and memory footprint 95%+.
  • Shipped bus thin-part detection: accuracy 60%→85%, inference ~4x faster, container size ~15x smaller.
  • Reduced model training compute cost 75%; training time from ~3 months to under a week.
  • Automated 3D synthetic-data generation (Blender, multi-GPU): 450+ day render workload → a week.
  • Engineered an interactive 360° car spin pipeline via novel-view-synthesis and multi-view geometry.
  • Prototyped generative-model approaches for background inpainting, reflection removal, and image harmonization.

Backend, Cloud & Platform

  • Drove a cost-reduction program: $64,000+/year in recurring savings.
  • Cut GPU inference hosting cost ~7x via self-hosted multi-model platform, two cloud migrations.
  • Led an auth overhaul cutting response time 99%+ to under 100ms.
  • Implemented a GCP AI inference gateway across 30+ endpoints with Redis rate-limiting + autoscaling.
  • Established MongoDB schemas and async task-queue infra across microservices.
  • Owned production incident response and DB reliability, including a retention cleanup that cut one major collection's size by a third.

Team Leadership & Delivery

  • Advanced from individual contributor to a technical lead role; mentored four AI and backend engineers while still shipping individual technical work end-to-end.
  • Owned the AI/CV technical roadmap and technical hiring; handled technical support escalations for major enterprise accounts.
  • Guided delivery of an automated social pipeline and internal analytics dashboards.
  • Steering an automated image quality-check and auto-correction system with AI-flagged human review routing.
2019 — 2020
Maintenance Engineer, Graduate Engineer Trainee
Reliance Industries Limited — Hazira, Gujarat · Mechanical plant maintenance at a large-scale manufacturing facility
// stack

Technical Stack

Tools and algorithms I use to ship production AI systems.

Languages & ML / CV
Python
JavaScript
PyTorch (multi-GPU, AMP/FP16)
OpenCV
Object Detection (YOLO)
Segmentation & Matting
GANs & Diffusion Models
Inference Acceleration (ONNX)
3D Reconstruction (NeRF, Gaussian Splatting)
AI/LLM & Agentic Systems
LLM APIs (Claude, OpenAI, Gemini)
LangGraph & LangChain
RAG & Agentic RAG
Multi-Agent Systems & Tool Calling
Model Context Protocol (MCP)
AWS (EC2, S3, Lambda, SageMaker)
GCP (Compute, Cloud Run, Vertex AI)
Docker & GitHub Actions
Self-Hosted GPU Inference Platforms
Backend, Data & Practices
MongoDB & MongoDB Atlas
PostgreSQL & pgvector
Redis & Async Task Queues (Celery)
FastAPI & REST API Design
Schema Design & Indexing
Git & Code Review
Production Incident Response
Technical Hiring & Mentorship
n8n & Playwright Automation
// currently building

Projects

🤖
In Progress

AI-Powered Job Application Agent

Aug 2026 — Present

Agentic system automating the end-to-end job application workflow: searches jobs, parses descriptions, matches candidate profile via RAG, generates tailored materials, and completes submissions through browser automation.

LangGraphRAGMCPFastAPIPlaywright
More Projects →
// credentials

Certifications

Key Coursera specializations most relevant to my current work.

Applied Data Science Specialization↗ Deep Learning Specialization↗ DeepLearning.AI TensorFlow Developer Specialization↗ IBM Applied AI Specialization↗ IBM Data Science Specialization↗ Reinforcement Learning Specialization↗
View All Certifications →
// background

About & Education

I started out as a Mechanical Engineer at VNIT Nagpur, then spent a year in plant maintenance at Reliance Industries before pivoting into programming and data science — completing a second Bachelor's degree at IIT Madras while building a career in applied AI from the ground up.

Since 2023 I've been at HelloRamp [HelloAR], where I progressed from individual contributor to technical lead, building the computer-vision systems, backend platform, and cloud infrastructure behind an AI visual merchandising product used by automotive marketplaces across three countries.

I care about production reliability as much as model accuracy — most of the work above is as much about cost, latency, and uptime as it is about pushing pixels through a neural network.

B.Sc., Programming and Data Science
Indian Institute of Technology (IIT), Madras · CGPA 9.2/10
2020 — 2023
B.Tech, Mechanical Engineering
Visvesvaraya National Institute of Technology (VNIT), Nagpur · CGPA 8.12/10
2015 — 2019

📄 Resume

View the full HTML version in-browser, or download the PDF — always up to date with my latest role and impact.

View Resume →

Let's talk

Open to interesting problems in AI, computer vision, and backend systems.