● ML & DATA SCIENCE — NEW DELHI, IN — PORTFOLIO / 2026

Smarpit Malik I build ML systems & full-stack products — LLM fine-tuning, RAG pipelines, deployed with the boring parts actually working.

$ whoami →
0repos
0hacks
0stars

About — the short version

(01)

I'm a developer working where machine learning meets shipping software that actually runs — currently deep in ML & data science, levelling up in deep learning and DSA.

My happy place is a hackathon: a vague problem statement, a team, a deadline, and a model that has to work by morning. That loop — prototype fast, measure, iterate — is how I went from QR-code generators to LLM-powered RAG applications.

PRAMANAM taught me the other half: auth, databases, deploys — the boring parts are what make software real. I like building them too.

Looking to collaborate on Python, ML, or data science projects. Ask me about Python, Pandas, or getting started with ML — I remember exactly what the first epoch felt like.

Tools I reach for

(02)
ML / Data
PythonPandasNumPyscikit-learnJupyterMatplotlib
Deep Learning / LLMs
PyTorchTensorFlowHugging FaceLangChainRAGFine-tuning
Languages
C++DartTypeScriptJavaScriptSQLBash
Web / Shipped
Next.jsPostgreSQLPrismaFlutterFastAPIHTML / CSSStreamlit
Currently learning
Deep LearningDSA in C++MLOpsTransformers

Selected work

(03) — INDEX
/001 PRAMANAMDocument & certificate verification — JWS-signed QR codes, PDF pipeline. Smart India Hackathon 2026, six-person team, live in production. ● liveTypeScript ↗ /002 LLM RAG ApplicationRetrieval-augmented generation — grounding LLM answers in your own documents instead of letting it guess. Python ↗ /003 Fine-tuning LLMsNotebooks on adapting large language models — dataset prep, training loops, evaluating what actually improved. Jupyter ↗ /004 NeuroSpark — Stage 1Hackathon build: an ML pipeline taken from raw data to working demo under a hard deadline. Python ↗ /005 Machine Learning LabCore ML algorithms implemented and stress-tested in notebooks — understanding what libraries do under the hood. Python ↗ /006 HackTheMindCross-platform Flutter app built with my team for a mental-health focused hackathon. Dart ↗ /007 The DSA ClimbMy public process of learning data structures & algorithms in C++ — every solved problem committed. C++ ↗

+ 19 more on github/smarpitm ↗

Eight hacks, seven months

(04)
Jul 2026

HackAgent AIx

AI agent system built end to end in Python with Team Coin Hustlers.

Jul 2026

NHAI Hackathon 7.0

Finalist — built the CV liveness module for India's national highway authority challenge.

Jul 2026

Bharatiya Antariksh

Space-tech problem solving for the national space hackathon.

Jun 2026

NeuroSpark

Raw data to working ML demo on a tight deadline.

Jun 2026

Hack2Skill

Rapid prototyping — shipped a functional Python build.

Apr 2026

HackTheMind

Cross-platform Flutter app for a mental-health problem space.

Mar 2026

Hack4Impact

Impact-first hacking aimed at real social outcomes.

Jan 2026

HACK4DELHI

Where the streak started — civic tech for Delhi.

avg. cadence: one hackathon every ~26 days · sleep schedule: deprecated

(05) — Contact / open to ML & DS collaborations

Let's train
something together