Kartik ShirodeResume

Available for projects

I build ML systems that ship.

LLM applications, vision pipelines, predictive dashboards, full AI web apps at fixed rates. Production work across NLP, computer vision, and self-supervised learning, written so the next person can build on it.

Hire MeView Work
Kartik Shirode
16projects built
$150fixed-rate projects start at
8.26CGPA, B.Tech CS

Projects

Things I've built. What they do, what worked, what didn't.

Featured

Tesseract (SIH 2026)

One drone pass in, a georeferenced, textured 3D model out, on one laptop GPU. A synthetic ten-minute clip ran in 548 s of a 900 s budget, at 0.344 m median error with 98.5% within 1 m of ground truth.

Built with
  • Python
  • COLMAP
  • OpenMVS
  • MapAnything
GitHubLive demo

Featured

UAV-X (PUSHPAK 2026)

Four-drone swarm, simulated in PX4 and Gazebo, that surveys a disaster zone beyond line of sight. Kill the relay mid-mission and the swarm elects a new one: route home back in 38.9 s, 120 of 120 cells covered, 5,917 of 5,917 packets delivered. Built for IIT Bombay Techfest's PUSHPAK Grand Challenge and selected for Stage 2, one of 14 teams from 448 registrations.

Built with
  • ROS 2
  • PX4
  • Gazebo
  • Python
Private / available on request

Featured

AnswerWala

Hindi and English voice question answering over 950K passages. Hybrid dense and BM25 retrieval, under 200 ms P90 from transcript to answer in both languages, and speculative decoding cut English generation p50 from 84.7 ms to 43.5 ms.

Built with
  • Python
  • FastAPI
  • vLLM
  • RAG
BlogGitHub

Featured

CallQA

Evaluation-first support call intelligence: ASR, diarization, and event detection benchmarked on 60 calls with exact ground-truth labels.

Built with
  • Python
  • faster-whisper
  • pyannote
  • Streamlit
Case studyGitHub

See all 16 projects →

Skills

The stack behind the work.

ML / AI
  • Python
  • PyTorch
  • TensorFlow
  • scikit-learn
  • HuggingFace
  • CatBoost
  • XGBoost
LLMs / NLP
  • Transformers
  • RAG pipelines
  • Hybrid retrieval (BM25 + dense)
  • FAISS
  • Embeddings
  • vLLM
  • Speculative decoding
  • Fine-tuning (LoRA / PEFT)
  • LLM-as-judge
Speech / Audio
  • faster-whisper
  • pyannote
  • WER / DER evaluation
  • TTS pipelines
Computer Vision
  • Multimodal architectures
  • 3D reconstruction (COLMAP, OpenMVS)
  • Deepfake detection
  • Object detection
  • ViT
Backend
  • FastAPI
  • Node.js
  • Express
  • Java EE
  • SQLite
  • REST APIs
Frontend
  • React
  • Next.js
  • TypeScript
Cloud / Infra
  • GCP
  • Cloud Run
  • Firebase
  • Docker
  • Slurm (HPC)
Robotics / Simulation
  • ROS 2
  • PX4
  • Gazebo
  • Drone swarms
Blockchain
  • Hyperledger Besu
  • Solidity
  • IPFS
Data / Analytics
  • Power BI
  • Tableau
  • SQL
  • Pandas
  • NumPy

About

A Computer Science student building practical ML.

I design and ship end-to-end machine learning projects across NLP, computer vision, and supervised learning. The focus is on measurable outcomes, mathematical clarity, and code other people can actually pick up and build on. Based in Pune, India, I work with clients remotely at fixed rates.

I learn by building complete projects rather than reading about them. Most of what is on this site started as a problem I wanted to understand, ended as something deployed with real metrics, and left me with a clear answer to “what would I do differently next time.”

B.Tech in Computer Science at Savitribai Phule Pune University, CGPA 8.26, class of 2028.

Experience

Where I've worked lately.

The two latest roles. The full history, including the Rajni Tech to Code NexIn rebrand, is on the experience page.

  1. Mar 2026 - Sep 2026

    ML & Research Intern

    MegaMinds IT Services

    Built production-leaning ML systems and applied-research prototypes for product work, across NLP, computer vision, and deep learning.

    Tools
    • Python
    • Deep Learning
    • NLP
    • Computer Vision
    • Applied Research

  2. Sep 2025 - Apr 2026

    Core Team Member (AI/ML)

    Google Developer Groups

    Drove the AI/ML track for the local GDG chapter. Built classroom-grade demos that explained real ML ideas to students new to the field.

    Tools
    • Python
    • scikit-learn
    • Supervised Learning
    • Jupyter
    • Workshops

See the full work history →

Contact

Drop a note.

Open to conversations about ML, data systems, and shipping real things. If it's role-specific, include the timeline and what you'd expect from me.

kartikshirode123@gmail.com

See the full contact page →