Large Language Models
Fine-tuning, quantization, local inference with Ollama & vLLM, context engineering, evals, and production RAG pipelines that actually work on messy data.
Now accepting students
Industry-grade Data Science, taught like a coach — not a lecturer.
Private 1-on-1 mentorship with a Lead Data Scientist at HP Inc. Learn exactly what's shipping in production today — LLMs, Multi-Agent Systems, RAG, MCP, and Quantitative Trading.
About your mentor
Built in production. Teaching from the trenches.
Kundan Singh Sorout
Lead Data Scientist · HP Inc.
Eleven years in engineering, seven of them in data science. I lead five engineers at HP Inc. building multi-agent systems over enterprise data warehouses, and before that shipped vision, diffusion and voice systems into retail, manufacturing and ecommerce. M.Tech in AI/ML from NIT Warangal; TensorFlow- and PyTorch-certified. The gap between what tutorials teach and what industry actually uses is enormous. I bridge it.
3
classes per week
30+
minutes per class
1 : 1
individual focus
100%
industry-grade content
What we build together
Four tracks. All grounded in what ships in production today.
Multi-Agent Systems
Agent architecture, Model Context Protocol (MCP), tool use, orchestration patterns, memory, and scaling multi-agent conversations without losing coherence.
Quantitative Trading
Feature engineering for time-series, deep reinforcement learning strategies, backtesting frameworks, and building signals that survive live markets — not just curves.
Industry Data Science
What's actually on your résumé vs. what companies want. Data pipelines, feature stores, MLOps basics, experiment design, and communicating results to non-technical stakeholders.
The method
Coach, not lecturer. You read first — I correct second.
Most courses explain a topic once, then expect you to code it. That's why people forget 80% by the next class. I flip it. You read the material at home. You explain it to me. I correct what's wrong and show the right mental model. Same cycle for code.
-
01
I assign
A focused topic or paper. One thing at a time. Expected reading: 30–60 minutes.
-
02
You read & absorb
You study it actively — take notes, form your own understanding, write your own explanation.
-
03
You explain back
Next class, you teach it back to me. The moment you stumble, we've found what to fix.
-
04
I correct & reframe
I patch the gaps, show the industry mental model, and connect it to what you already know.
-
05
You implement
You write the code. I review. Same loop — you attempt, I correct, you refine until it's production-grade.
Why this works for working professionals
This isn't about the money. It's the kick of changing one career at a time.
8–10 focused hours/month on you alone
A live cohort class splits the instructor across 30+ students. You get the full hour, every session.
Your pace, not the syllabus's
Serious students finish fast. Others go deeper. No artificial deadlines — only real skill gained.
Targeted at exactly your gaps
Strong in Python but weak at system design? We spend 80% there. No wasted classes on things you already know.
Students
Two people who changed track.
Both students are named, with a link to their own LinkedIn profiles, because a testimonial you cannot check is worth nothing.
When I started working with Kundan, I was an Angular developer.
What really changed my way of working was his approach — first read and understand the material, and then explain it back to him. This helped me quickly identify what I truly understood and what I only understood partially.
Today, I work on building GenAI and LLM applications.
Going from teaching mathematics to analytics looked impossible from the outside. What made it work was being corrected constantly — I would explain something, and he would show me exactly where the reasoning broke. That is the part no course does for you.
Pricing
Simple. Monthly. No lock-in.
Choose 1:1 or share the experience with a friend. Either way, you get the same depth.
Solo
1-on-1 Mentorship
Full attention. Built entirely around your gaps, goals and speed.
₹
15,000
/ month
- 3 private classes per week
- 30+ minutes each — no hard cap
- Reading + implementation review
- Curated resources & code reviews
- Pace set entirely by you
Most popular
Duo
Batch of Two
Bring a friend or colleague. Same depth, split cost. Max 2 students per batch.
₹
10,000
/ student / month
- All Solo benefits included
- Max 2 students — never more
- Learn with accountability partner
- Peer-explain builds depth faster
- Same ₹10,000 threshold for both
No contracts. Month-to-month. Pause whenever life gets busy — we pick up where we left off.
Questions
What people ask before they apply.
If the answer you need is not here, ask it in the form below — it is the same inbox either way.
Who is this for?
Working professionals who already write code and want to close a specific gap - not beginners looking for a first syllabus. The plan is built around your gaps, your goals and your speed. Strong in Python but weak at system design? We spend eighty per cent of the time there, and none on what you already know.
How much time does it take each week?
Three private classes a week, thirty minutes or more each with no hard cap - eight to ten focused hours a month on you alone. Between classes, expect thirty to sixty minutes of reading: you study the material first, then explain it back.
What do we actually cover?
Four tracks, all grounded in what ships in production today. Large Language Models - fine-tuning, quantization, local inference with Ollama and vLLM, context engineering, evals and production RAG. Multi-Agent Systems - agent architecture, Model Context Protocol, tool use, orchestration and memory. Quantitative Trading - time-series feature engineering, deep reinforcement learning, backtesting and signals that survive live markets. Industry Data Science - pipelines, feature stores, MLOps, experiment design and communicating results to non-technical stakeholders.
What does it cost, and am I locked in?
₹15,000 a month one-on-one, or ₹10,000 per student a month if you bring someone. No contracts and no lock-in: it is month to month, and you can pause when life gets busy and pick up where you left off.
Can I learn with a friend or colleague?
Yes. The batch of two is ₹10,000 per student per month and is capped at two students, never more. Same depth as one-on-one, and having to explain things to a peer tends to build that depth faster.
How is this different from a cohort course?
A live cohort splits one instructor across thirty or more students. Here the session is yours, at your pace rather than a syllabus's, with no artificial deadlines. The method is inverted too: most courses explain a topic once and then expect you to code it, which is why people forget most of it by the next class. You read first, explain it back, and I correct the mental model before you implement.
Who is teaching?
Kundan Singh Sorout, Lead Data Scientist at HP Inc., leading five engineers building multi-agent systems over enterprise data warehouses. Eleven years in engineering, seven of them in data science, with vision, diffusion and voice systems shipped into retail, manufacturing and ecommerce before that. M.Tech in AI/ML from NIT Warangal, and TensorFlow- and PyTorch-certified.
How do I start?
Fill in the application form on this page. If there is a fit you will hear back within forty-eight hours, and there is a free twenty-minute intro call before you commit to anything. I only take on students I can genuinely help.
Apply
Tell me about you.
I only take on students I can genuinely help. Fill this out — if there's a fit, you'll hear from me within 48 hours with next steps, a short call, and a plan.
- ✓ Replies within 48 hours
- ✓ No spam, ever
- ✓ Free 20-minute intro call if we're a fit