The Builder
“Consistent output, genuine execution track record.”
AI Engineer @ The Product Folks | Building Intelligent Products | Generative AI • LLMs • AI Agents • Computer Vision
Hey, I'm Zalak.I'm an AI Engineer but AI is only one part of what I'm interested in.I like understanding how things work, taking ideas apart and then trying to build something with them. That curiosity has naturally led me toward AI especially Generative AI, LLMs, AI Agents, Machine Learning and Computer Vision.What I enjoy most is the space between an idea and something that actually works. Sometimes that means experimenting with a model, sometimes writing the software around it and sometimes f
In their voice · Build Story
I keep shipping systems that do one thing well: handle volume without breaking.
I started with an AI content pipeline at The Product Folks that generated 200+ branded assets. The hard part wasn't getting the AI to generate content. It was making the workflow reliable enough that one failure didn't stop everything. I automated the repetitive steps and structured the pipeline to handle failures without blocking the rest.
Then I built a semantic search engine across 23K+ talent profiles with sub-1-second query latency. The temptation was to make the retrieval system more sophisticated. Instead I focused on efficient indexing and keeping the search path simple. That was enough to get the performance we needed.
I took the same approach with my sign language detection model which reached 98.78% accuracy. The hardest part was improving data quality and iterating on the model. I built that alone because I wanted to understand the entire path from data preparation to training and evaluation rather than only owning one part of it.
That mindset also shaped Gitlane. I shipped it as both a CLI and VS Code extension because the challenge wasn't just making the AI work. It was making the tool fit naturally into a developer's existing workflow.
The pattern I've noticed across these projects is that good engineering isn't always about building the most sophisticated system. It's about knowing what needs to be reliable, keeping the right things simple and shipping something people can actually use.
Skills
Artificial Intelligence (AI)Generative AIMachine Learning
Journey · shipped & built
2026shipped
Designed and shipped an AI content production pipeline via Higgsfield MCP that converts plain-text briefs into ready-to-publish videos and images. Generated 200+ branded creatives for community campaigns and marketing at a fraction of the time and cost of a manual creative pipeline.
2026shipped
Built an AI candidate-matching platform that ranks best-fit profiles from a 23,000+ indexed professional pool against any uploaded job description, turning multi-day manual screening into a seconds-long semantic search. Deployed as a production Next.js app delivering top-ranked candidate lists in under 1 second per query.
2026shipped
Delivered Arova, a production conversational AI platform powering 24/7 community engagement across 8-10 high-frequency intent flows, cutting response latency from hours to under 2 seconds and eliminating 80% of manual triage effort.
For founders
You're seeing the public profile. Founders see the full read — honest strengths, where this person won't fit, and the three questions to ask on the first call.