
L1 · About
hello there! ↘
About me, tokenised.
Tap or hover an underlined word to see what it attends to.
I got here the unglamorous way—one curious question about how something worked leading to a lot of broken code and a few systems that eventually didn't break. No grand plan, just a lot of iteration. Somewhere between fixing one bug and designing the next feature, code stopped being an assignment and became the way I think through problems.
What I care about is whether the thing actually runs—whether the model generalizes, whether the system scales, whether the code someone else touches makes sense six months later. I'm still learning most of this as I go, and I've made peace with that. There's always a harder problem worth chasing, and a better version of the last system worth building.
View the magic, documented.“Excellence is a continuous process and not an accident.”
↻ hover or tap to run style transfer
L2 · Projects
explore my work! ↘
The case files.
My projects, bound as a dossier. Turn the pages (scroll, or swipe on phones) or jump straight to one with a bookmark.
page 1 / 5
L3 · Experience
where i've worked ↘
Experience.
Where I've shipped real systems for real teams.
−90%repeat-query latencyTech Intern @NTPC Vindhyachal
May – June 2026Madhya Pradesh
- ▹Built a Gemini 2.5 Flash & sentence-transformer RAG pipeline to automate support ticket resolution and email notifications.
- ▹Cut repeat-query response latency by 90% (from ~3s to ~300ms) by implementing semantic caching and confidence-scoring filters.
- ▹Established a self-improving flow: resolved tickets are auto-embedded and indexed, allowing the chatbot to answer future queries without human intervention.
Next.jsTypeScriptMongoDB AtlasSentence TransformersGeminiFastAPIRAG
Data Analytics Intern @Pegasus Mortgage
May – June 2026Remote
- ▹Developed a Python-based broker analytics pipeline using Pandas and scikit-learn for data cleaning, feature engineering, and identifying patterns in broker activity.
- ▹Performed exploratory data analysis and statistical analysis to uncover broker performance trends, segment activity patterns, and generate actionable insights from mortgage datasets.
PythonPandasscikit-learnData AnalyticsFeature Engineering
L5 · Education
where i learned ↘
Education & certifications.
Checkpoints saved along the way.
B.Tech. in Computer Science (Specialization in AI/ML)
Vellore Institute of Technology, Vellore, Tamil Nadu
CGPA: 9.26
degree in progress…
Class XII (CBSE)
Delhi Public School, Ranchi, Jharkhand
Percentage: 93.6%
Class X (CBSE)
Delhi Public School, Ranchi, Jharkhand
Percentage: 94%
Class X (CBSE)
Delhi Public School, Ranchi, Jharkhand
Percentage: 94%
Class XII (CBSE)
Delhi Public School, Ranchi, Jharkhand
Percentage: 93.6%
B.Tech. in Computer Science (Specialization in AI/ML)
Vellore Institute of Technology, Vellore, Tamil Nadu
CGPA: 9.26
degree in progress…
Certifications & Training
Forage Virtual Experience Programs
- British Airways (Data Science)
- Deloitte (Data Analytics)
- Quantium (Commercial Analytics)
- Tata Group (GenAI)
Forage Virtual Experience Programs
- British Airways (Data Science)
- Deloitte (Data Analytics)
- Quantium (Commercial Analytics)
- Tata Group (GenAI)
L6 · Stack
my toolbox ↘
Tech stack.
The tools I reach for.
L7 · Contact
come say hi! ↘
Let's build something intelligent.
Tap a building and mini-me plans a route with A*, walks over and rings the bell. Tap anywhere to roam, use the arrow keys, or tap mini-me to say hi.

















