Cloud and AI/ML, taught by someone who runs it in production.
I'm Pushpjeet Cholkar. For 17 years I've built, tuned and rescued Oracle, Linux and AWS systems for Deutsche Bank, IBM, Wipro, Accenture, Capgemini, Cognizant and Oracle University — across four continents. Now I train the engineers who run those systems, and consult on the AI layer going on top of them.
Pushpjeet CholkarCloud & AI/ML Consultant · Corporate Trainer Indore, India — working worldwideAvailable for work
pushpjeet@pitc — sqlplus
SQL> SELECT metric, value FROM pitc.track_record;
METRICVALUE-------------------------- ------------------------------
Years in industry 17
Enterprise clients 40+
Countries delivered in 12+
Oracle releases run 8i → 26ai
Cloud AWS SAA · Oracle Cloud Infra
Engineers trained 3,000+
Training feedback 100% positive
Current focus AI/ML on the data layer
8 rows selected.SQL>-- next: your team, your stackSQL>_
17Years
40+Clients
12+Countries
3,000+Trained
100%Feedback
8i→26aiOracle
What I do
Three ways to work with me
The same engineering experience, delivered at three different sizes — a company, a team, or one person deciding to level up.
Consulting
For companies
Someone senior who has already fixed this before. Performance tuning, migrations, cloud architecture, and getting AI/ML working against real production data.
Custom cohorts built around your stack, not a generic syllabus. Delivered on-site or live-virtual, the same way I've delivered for Oracle University and AWS reStart.
Small public cohorts for students and working professionals. Some paid, some free. Live sessions, real labs, recordings, and a certificate you actually earned.
New AWS batches, AI/ML tutorials, Oracle tips and blog posts. No spam, unsubscribe anytime.
University lab session
Since 2009
Three thousand engineers, one room at a time
Corporate floors in Pune and Bangalore, university labs, bootcamp cohorts in Germany and the UK, and Live Virtual Classes for Oracle University reaching people I never met in person. Same approach every time: short explanation, then hands on the keyboard.
Track record
Where the work has been
Consulting and training engagements delivered directly or through partners, from Pune and Bangalore to Cape Verde, Muscat, Singapore, Suva and Berlin.
Deutsche BankOracle UniversityOracle CorporationIBMWiproAccentureCognizantCapgeminiTech MahindraAmdocsMphasisAWS reStart UKAWS reStart GermanyGovernment of FijiIndian Air Force (via NIIT)SriLankan AirlinesATC MuscatNOSI Cape VerdeC-DACSynechronIGATEJDAAris GlobalZeOmegaNobleProgEdurekaneue fischeParul UniversityITM University
100% positive feedback across corporate training engagements, and Live Virtual Classes delivered worldwide for Oracle University, Singapore.
Free for everyone
Labs, projects and material — no charge
Teaching has always been the part I enjoy most, so a lot of it is simply open. Build the lab, break it, ask me questions.
Oracle 23ai on Docker
Lab
Stand up a multitenant container database on your laptop, create PDBs, and practise DBA work without touching a licence server.
AWS VPC from scratch
Lab
Build a three-tier VPC by hand — subnets, route tables, NAT, security groups — then tear it down. The lab that makes the SAA exam click.
Build a RAG tutor
Project
A retrieval-augmented tutor over your own PDFs, end to end: chunking, embeddings, vector store, evaluation. My live research area.
A slow database, a cloud migration nobody wants to own, a team that needs to be trained properly, or a batch you'd like to join. WhatsApp is fastest — I usually reply the same day.
WhatsApp +91 70492 35525
Email pc@pushpjeet.com
pushpjeet@gmail.com
LinkedIn /in/pushpjeet-cholkar
Instagram @coachpushpjeet
Office 8/5 Manas Mayfair, 4th Floor
South Tukoganj, Indore 452001
Entities PITC Solutions LLC (USA)
Pushpjeet IT Consultant (India)
Get the Most, from the Best!!
02 — Training & Batches
Live batches, not recorded courses
Every batch is taught live by me. You get the labs, the recordings, and the ability to interrupt and ask why. Batch sizes stay small enough that I know who is stuck.
AWS session — Amazon FSx, live in the console
AWS S3 in Detail — From Zero to Pro
Starts 18 Oct
One service, all the way down. Buckets and objects, storage classes and lifecycle rules, versioning, encryption, bucket policies versus IAM, replication, static hosting, event notifications, and the cost traps that surprise people on their first bill.
The SAA-C03 blueprint taught through labs, not slides. VPC, IAM, S3, EC2, RDS, ELB, CloudFront, well-architected trade-offs. Built from three AWS reStart cohorts.
The Linux a DBA or cloud engineer actually uses: filesystem, permissions, processes, networking, shell scripting, log hunting. Run free, a few times a year.
Reading execution plans, indexing strategy, statistics, hints, and the twelve query shapes that cause most production slowness. Bring your own slow query.
Build a working retrieval-augmented application over real documents: chunking, embeddings, vector stores, prompt design, evaluation, and what breaks in production.
Infrastructure as code from first principles — providers, state, modules, plan and apply discipline. The curriculum I deliver for neue fische × SPICED.
Don't see your topic? I also teach MongoDB, PostgreSQL, MS SQL Server, Oracle Forms & Reports, data warehousing, XML technologies, Java, MATLAB and Unix essentials. Ask and I'll tell you honestly whether I'm the right person for it.
Inside the room
What a session actually looks like
University labs, corporate floors, and my own training room in Indore. Small explanation, then hands on the keyboard — which is why these photographs are mostly of people working rather than of me talking.
Full lab, university programmeCohort in sessionHands-on — everyone on a machineOracle programme cohortCorporate session, on the floorPITC training room, IndoreAddressing a full hallCorporate cohortBatch complete
For organisations
Corporate training
Most of my seventeen years has been spent training engineering teams inside large companies. The syllabus gets written after I understand your stack, not before.
How a corporate engagement runs
Scoping call — your stack, your team's current level, the business reason for the training.
Custom curriculum — written to your systems, reviewed by you before delivery.
Delivery — on-site, live-virtual, or blended, in your time zone.
Free AWS material on this site: the S3 vs EBS vs EFS storage guide, AWS in the AI/ML tutorials (Amazon Bedrock, SageMaker AI), Linux and Bash for labs, and a free SAA study plan.
Format
A hub page that lists every AWS guide, tutorial and training page on this site
Start here
S3 vs EBS vs EFS: object, block and file storage compared, with current limits and AWS CLI steps
Also here
Amazon Bedrock in the RAG tutorial, SageMaker AI in the ML Solutions Quest, and the free SAA 30-day study plan
Learn Bash scripting from zero to pro: 45 modules and 162 tested scripts with real output, exercises, strict mode, trap, getopts, cron and a capstone tool.
Format
45 modules in five levels plus a server-health capstone; every example shows the script, the command and its real output
Practice
25 exercises with worked solutions, a 25-question quiz and all 162 scripts in one zip
Multiple linear regression in Python at expert depth: OLS normal equations, a worked example, VIF and Cook’s diagnostics, Ridge vs Lasso, and three labs.
Format
Expert written tutorial with a worked example by hand and full runnable code
Practice
3 Python labs, a 10-question expert quiz and a cheat sheet with 15 flashcards
Covers
OLS in matrix form, adjusted R², VIF, Cook’s distance, Breusch–Pagan, Ridge, Lasso, Elastic Net
From “What is AI?” to foundation models, RAG and the AWS AI services — no background in coding or maths needed. Six modules that build on each other: AI vs ML vs DL vs GenAI, ML fundamentals, deep learning, generative AI, and Amazon Bedrock vs SageMaker AI.
Interactive
Reveal-the-answer questions, a scored quiz, flashcards and a searchable glossary
PDF guide
62 pages · 88 questions answered from basic to advanced · 30 quiz questions · 32 practice challenges
How machines learn to create new images. What latent space is, how a Variational Autoencoder builds a smooth one, how a GAN's generator and discriminator compete, and where both show up inside today's image generators.
Format
Video lesson plus written tutorial with diagrams
Practice
8 self-check questions and a hands-on VAE challenge on MNIST
A gamified warm-up for the AI & ML Fundamentals tutorial. Eight short levels with mini-games, from the AI family and how models learn to generative AI, RAG and the AWS toolbox. Earn XP and badges as you go.
Format
8 levels with mini-games, checkpoints, a mind map, flashcards and a boss battle
Practice
Train a model, build a neuron, be an LLM, estimate an LLM bill
What an AI agent is and how it works: the brain, tools, memory and planning. Watch an agent run its think–act–observe loop, learn when a human must approve, and meet Amazon Bedrock Agents, AgentCore and MCP.
Format
Bonus level with an agent-loop simulator, flip cards and two games
Practice
Pick the right tool, play safety officer, clear the checkpoint
A gamified warm-up on developing machine learning solutions. Take one model through its whole life, the way Amazon did for its call centre: business goal, data, training, evaluation, deployment and MLOps with Amazon SageMaker AI. Earn XP and badges as you go.
Format
8 levels with mini-games, checkpoints, a mind map, flashcards, myth or fact and a boss battle
Practice
Order the ML lifecycle, split your data, diagnose a model, choose the metric, pick the deployment option
The Goldilocks Problem: the bias-variance trade-off
Video
How does an AI know when to stop learning? A short video on underfitting, overfitting and finding the balance, with a scene-by-scene walkthrough that tests each model on brand-new data.
Format
74-second video plus an interactive scene-by-scene walkthrough
Practice
Model doctor and prescribe-the-cure games, plus a checkpoint
From “what is rwx?” to setuid, sticky bits and umask. Read the levels, play the mini-games, hack the practice terminal, earn XP and badges, and finish as a Permissions Root.
Format
16 levels with mini-games, a sandbox terminal with missions, an Octal Dojo and an FAQ
Practice
6 hands-on labs, a 20-question MCQ challenge and practice questions with model answers
Covers
rwx, owner/group/other, ls -l, chmod, chown, chgrp, sudo, special bits, umask
Processes, ps, pstree, top, kill, nice, jobs, at and cron. Pick a path, and every topic shows just that level — switch level on any topic whenever you want to go simpler or deeper.
Format
Beginner, intermediate and expert paths across 7 topics, with a simulated process-lab terminal
Practice
8 hands-on labs, 30 MCQs (10 per level), FAQ and practice questions with model answers
Covers
PIDs, process states, signals, nice/renice, job control, at, cron
Docker Compose to a running CDB, then create, clone, unplug and plug PDBs. Includes the teardown script so you can start clean.
DockerOracle 23ai2–3 hrs
AWS VPC from scratch
Lab
A three-tier VPC built by hand in the console: public and private subnets, route tables, internet and NAT gateways, security groups, bastion access.
AWSNetworking3 hrs
Terraform: build and destroy
Lab
The same VPC, this time as code. Providers, variables, modules, remote state, and the discipline of reading a plan before you apply it.
TerraformIaC3 hrs
Linux for DBAs: 40 drills
Lab
Forty small command-line exercises with answers — permissions, processes, disk, logs, networking, and the shell one-liners you'll use weekly.
LinuxShellSelf-paced
Build a RAG tutor
Project
A retrieval-augmented tutoring assistant over your own PDFs: chunking strategy, embeddings, vector store, retrieval evaluation, and honest failure analysis.
PythonLLMResearch area
Superset on Parquet
Project
Containerised analytics stack: a historical Parquet dataset loaded and visualised in Apache Superset via Docker. A genuine portfolio piece.
SupersetParquetDocker
Twelve slow queries
Lab
Twelve deliberately bad queries with their execution plans. Fix each one, then compare with my fix and the reasoning behind it.
SQLTuningAdvanced
ETL pipeline starter
Project
Extract from a source database, transform with Python, load to a warehouse — with logging, idempotency and restartability built in from the start.
PythonETLData eng
Database health-check pack
Scripts
The SQL scripts I actually run on day one of a tuning engagement: tablespace pressure, top waits, missing indexes, stale statistics, session activity.
OracleDBAScripts
Reference material
Downloads
Cheat sheets and handbooks, free. Ask for any of them and I'll send the file — that way I also know what people actually need more of.
AI & ML Fundamentals — student guideSix modules from AI basics to RAG and AWS AI services, with quizzes and practice challenges · 62 pages · PDF
Companies usually call me for one of four reasons: something is slow, something has to move, something needs designing properly, or someone wants AI working against data they already have.
Something is slow
Performance
Database and SQL performance tuning on Oracle, PostgreSQL, MS SQL and Snowflake. Execution plans, indexing, statistics, contention, storage. I've done this for a national ERP in Cape Verde and for banking systems in Pune.
Health check and prioritised findings, usually within a week
Query-level tuning with before/after measurement
Instance and storage-level configuration review
Knowledge transfer so the fix doesn't leave with me
Something has to move
Migration
Version upgrades across Oracle 8i through 26ai, cross-platform migrations (Oracle ↔ MySQL, PostgreSQL), and lift-and-shift or re-architecture onto AWS and Oracle Cloud Infrastructure.
Assessment, risk register and rollback plan before anything moves
Multitenant conversion and PDB strategy
Cutover rehearsal and go-live support
Post-migration tuning, which is where most projects go wrong
Something needs designing
Architecture
Data warehouse and ETL design, high availability and replication, backup and recovery strategy, and AWS architecture reviewed against the well-architected framework.
Warehouse and dimensional modelling
ETL pipeline design with restartability built in
HA, replication, RMAN backup and tested recovery
VPC, IAM and cost architecture on AWS / OCI
AI on your own data
AI / ML
The layer everyone is being asked to add this year. Retrieval-augmented generation over internal documents, ML models against operational data, and a realistic assessment of what will and won't work.
RAG systems over enterprise documents
Machine learning on operational and warehouse data
Evaluation — how you'll know it's actually working
An honest opinion on whether AI is the right answer at all
Working together
Engagement models
DiagnosticA few days. I look at the system, and you get a written findings document with prioritised, costed recommendations. Often the whole engagement.
3–5 days
ProjectA defined outcome with a defined end — a migration, a tuning programme, a warehouse build. Fixed scope, agreed milestones.
4–16 weeks
Retained advisorA set number of days each month for design review, escalations and mentoring your team. The model that works best for growing engineering groups.
Monthly
Train and hand overConsulting and training combined: I fix it, then teach your team how it was fixed so the capability stays in-house.
Blended
Contracting
Two entities, so clients can contract on whichever side suits them:
USA
PITC Solutions LLC
India
Pushpjeet IT Consultant, Indore
Remote
Worldwide, any time zone
On-site
India and international, by arrangement
Start with a conversation
Tell me the symptom, not the solution. A twenty-minute call is usually enough for me to say whether I'm the right person, and what I'd do first. No charge for that.
I started as an Oracle developer and DBA in 2009, and spent the first decade doing the unglamorous work — tuning queries that had been slow for years, recovering databases at two in the morning, migrating systems that nobody had documented.
Somewhere in there I discovered I was better at explaining it than most people were, and training took over. I've since delivered for Oracle University, AWS reStart, and engineering teams at IBM, Wipro, Accenture, Capgemini, Cognizant and Deutsche Bank, in India and across a dozen other countries.
Today the work splits three ways: consulting on databases and cloud, training engineers and university students, and research into how AI can actually be used to teach — which is where I intend to spend the next stage of my career.
Seventeen years only counts if the last year is in it
The stack I teach keeps moving, so I keep showing up where it moves — cloud-native conferences, new Oracle releases the month they land, and the AI tooling that is rewriting how all of this gets built. What I teach on Monday is what I was reading about on Sunday.
Research
What I'm working on academically
I'm working towards a Ph.D. with a target intake of Fall 2028. The research question I keep returning to is whether a language model can genuinely teach — and how you'd measure it if it could.
Primary interests
RAG-based tutoring systems — retrieval-augmented models as subject tutors, and where the retrieval layer fails students
Learning analytics — measuring what a learner has actually understood, not what they clicked
Evaluation of AI tutors — designing benchmarks that reflect learning outcomes rather than answer similarity
Query optimisation — carried over from seventeen years of tuning work
Teaching & supervision interests
Database systems — Oracle, SQL, PL/SQL, PostgreSQL, MongoDB
Data structures and algorithms
Operating systems — Unix and Linux
Cloud computing — AWS and Oracle Cloud
Data warehousing and business intelligence
Artificial intelligence and machine learning
Software engineering and project management
Open to research collaboration, co-authorship, and guest lectures. If you're working on AI in education or learning analytics, I'd like to hear from you.
06 — Connect
Let's talk
Fill this in and it becomes a clear message you can send to me in one tap — by WhatsApp or by email, whichever you prefer. Nothing is stored on this page.
India office8/5 Manas Mayfair, Fourth Floor, opp. Nath Mandir Road, South Tukoganj, Indore, Madhya Pradesh 452001
EntitiesPITC Solutions LLC — USA Pushpjeet IT Consultant — India
Response time
WhatsApp is fastest — usually the same day. Email within two working days. For consulting enquiries, say what system and what symptom and I can be useful in the first reply.
pushpjeet.com is the personal website of Pushpjeet Cholkar (Pushpjeet IT Consultant). Privacy contact: pc@pushpjeet.com.
The newsletter stores your email address, optional first name and topic choices, and only after you confirm by clicking the link in the confirmation email (double opt-in).
Emails are sent with Amazon SES from news@pushpjeet.com. Every email has an unsubscribe link, and you can unsubscribe at any time.
No selling or renting of personal data, no advertising cookies, no analytics trackers.
