Kunal kasodekar

Hey there! I'm Kunal a Founding AI Engineer at a stealth startup, where I’m leveraging GenAI to automate marketing workflows.I have a MS in Computer Science from ASU.I live and breathe Computer Vision and Deep Learning, turning complex problems into elegant solutions.I’m all about building end-to-end ML pipelines, mastering every step from data cleaning and feature engineering to modeling, training, deployment, and monitoring. My projects have taken me from detecting bark beetle infestations to satellite surveillance systems and beyond.Why do I do this? Because I love it. I’m passionate about creating cutting-edge solutions that push the boundaries of what’s possible. Whether it’s optimizing models for real-time applications or diving deep into neural networks, I’m always up for a challenge.Take a look around to see some of the cool stuff I’ve been working on. If you’re as excited about machine learning as I am or have an awesome project in mind, let’s connect!

I am passionate about deep learning, Generative AI, and creating end-to-end ML pipelines. I am deeply captivated by the fields of Multimodal Learning, NLP, and CV, deriving immense pleasure from building models from the ground up, working with images and text, and crafting solutions for real-life problems. I possess a deep understanding of Transformer and Vision models and experience writing and training models from scratch in PyTorch. Some of my relevant work in the area includes:

AI Engineer
Stealth Startup
(10/2024 - Present)
Remote, USA• Working on multi-agent LLM systems for Marketing.
• Generated brand-compliant display ads and Linkedin Single Image Ads using GenAI workflows tailored for B2B marketing clients, ensuring branding consistency and content relevance whilst reducing ad-creation time by 95%.
• Working on evaluating generated content using LLM as a Judge.
• Worked on Knowledge center population using using Contextual RAG and Hybrid Search.
• Completed migration for the backend from Django to FastAPI with SQLAlchemy and Alembic.
• Migrated to celery as our job queue/orchestrater from timeloop.
• Experience in designing and creating e2e GenAI workflows, LLM Multiplexing, LLMs, Tracing, Benchmarking, Optimization, content-refinement, Prompt Engineering, and testing.

Visiting Researcher
Kerner Lab
(08/2023 - 09/2024)
Tempe, Arizona• Developed multi-modal segmentation models for Mars surface analysisResearch Aide
WP. Carrey
(08/2023 - 01/2024)
Tempe, Arizona• Developed and deployed low-latency, optimized models for gun detection in CCTV streams

SWE Intern
Rocket Central
(05/2023 - 08/2023)
Detroit, Michigan• Created and deployed a web crawler with Terraform IAC modules for GCS results, including a search bar, results page, and NestJS backend.

Senior Software Engineer
Wipro Limited (CTO Office)
(07/2019 - 06/2022)
Bengaluru, India• Detected bark-beetle infestations and conducted pixel-wise segmentation using DenseNet and UNet.• Developed various solutions, including time-series forecasting, meeting minute generation, and mispacked medical item detection, while leading an ML/DL team and utilizing advanced model explainability techniques.

I'd love to hear from you! Whether you have a project in mind, need some advice, or just want to chat about all things machine learning, feel free to reach out.