ANANTJAIN
an education, eight shipped builds, a wall of skills and a shelf of trophies, all of it lives on this one page. keep scrolling, the particles know the way.
the engineer
behind the code
AI & Data Science undergraduate with a bias toward production software. Instead of keeping models confined to notebooks, I engineer pipelines that turn messy real-world data into measurable operational impact.
Whether predicting freight turnaround bottlenecks, designing performant neural microservices, or building desktop enterprise software from scratch, I treat software engineering and machine learning as two halves of the same discipline. Code must be clean, algorithms must be lean, and systems must solve real operational problems.
Predictive modeling, deep learning pipelines, statistical machine learning, and data-driven forecasting on real messy datasets.
Algorithmic complexity optimization, high-performance backends, low-latency microservices, and robust data structures.
Transforming analytical models into reliable, production-grade applications, robust APIs, and enterprise desktop tools.
{
"engineer": "Anant Jain",
"core_focus": ["Applied AI/ML", "Systems Engineering", "Algorithms"],
"obsession": "cutting latency & computational bottlenecks",
"location": "Delhi · Open to Relocate",
"availability": "Open to High-Impact Internships"
}the
education
three and a half years of artificial intelligence & data science at gati shakti vishwavidyalaya, with a transportation & logistics lens on all of it.
the missing 1.41 cgpa was mostly spent debugging.
the
experience
swipe the deck. three internships: real software, real data, real railways. no filler.
parksight
bengaluru illegal parking intelligence platform: a full-stack operational dashboard over 298,000+ parking violations across 802 geohash zones, with real-time risk prediction and patrol shift planning. an xgboost poisson regression under the hood, served via fastapi on render + vercel.
patrol shift planning off a poisson regression, dashboards people actually open, and one city's parking chaos turned into numbers that argue for you.
foliogpt
a shakespearean next-word predictor: type a line, get the next word, except the model believes it's 1595. lstm, gru, 1d cnn and distilgpt2, trained on 202,000+ words of the bard.
four architectures, sparse tokenization, and a vram reduction that sounds like a typo. served via flask + docker on hugging face spaces.
the real one runs on hugging face. go argue with the bard yourself.
the
toolkit
every bubble in that cloud has shipped something real. the "learning" list is longer and more embarrassing.
the shelf
drag through the trophy shelf. hackathons, ranks, and one domain that let me run the workshops for a change.
most people bounce at the top, you made it to the bottom.
final-year ai & ds student. competitive programmer. professional debugger of my own code. got into all this out of plain curiosity, still chasing the project that re-sparks it.
... followers
career milestones
