Om Pilaji
Co-founder · Director · Product and client engagement
Decides what the platform does next, and is the person a client talks to from the first conversation through to handover.
Technology consulting · Software engineering · Built in India, working worldwide
Engineering intelligent systems for complex problems.
Kalman Consultancy Services designs and builds software systems that turn operational data into decisions — combining software engineering, applied machine learning and data infrastructure. Based in India, working with organisations wherever they are.
Scroll
01 — About
Technology built around the problem.
Kalman is a technology consulting and software engineering company. We take a specific operational problem — returns, workflow, reporting, a manual decision made a hundred times a week — and build the system that solves it.
The name is the algorithm. A Kalman filter takes noisy measurements and produces an estimate you can act on. That is the pattern behind our engineering: not more dashboards, but systems that decide.
We are based in Hyderabad, India, and the work is remote by default. Returns cost the same whether the warehouse is in Pune or Rotterdam, so we take enquiries from anywhere and quote in your currency. We have no international clients yet — when that changes, this paragraph will say so.
Sarvajeet Bajikar
Co-founder · Director · Engineering and architecture
Owns the data model, the services and the decision engine that turns a graded item into a route.
Where we are
2026
Kalman founded by Om Pilaji and Sarvajeet Bajikar
Done
2026
ReturnIQ enters active development
In progress
August 2026
Company name cleared at the Ministry of Corporate Affairs
Done
Next
Incorporation as a Private Limited Company — SPICe+ filing with our Chartered Accountant
In progress
Then
First paying engagement, and the platform running with a first pilot customer
Planned
02 — Capabilities
What we build.
AI / ML
Applied machine learning
Classification, extraction, forecasting and ranking, put inside a workflow where they change a decision. We will tell you when a rule set would work better than a model.
SOFTWARE
Software engineering
Backend services, APIs and web applications built to a written brief and handed over with the source and documentation.
ENTERPRISE
Operational platforms
Systems for the parts of a business that run on spreadsheets — intake, grading, routing, approval, reporting.
DATA
Data engineering
Getting operational data out of email threads and spreadsheets into something queryable, then keeping it current.
VISION
Computer vision
Webcam and camera input turned into usable signal — tracking, gesture, condition assessment — with filtering that makes the output stable enough to act on.
AUTOMATION
Automation
Removing the steps a person repeats by hand every week, and the errors that come with them.
CLOUD
Infrastructure
Deployment, environments and monitoring for the systems we build, sized for the organisation that has to run them afterwards.
ARCHITECTURE
System architecture
Data models, service boundaries and integration design, reviewed before code is written rather than after.
R & D
Research and prototyping
Building the smallest version that answers whether an idea works, before committing budget to the full system.
ADVANCED SYSTEMS & RESEARCH
Advanced systems and research
Autonomous multi-agent simulation, semiconductor design and verification tooling, space and satellite software, and financial infrastructure. These are long-term engineering interests rather than services we sell today — they sit on the roadmap, and we say so plainly rather than listing them as capabilities we can deliver this quarter.
03 — Services
From idea to engineered system.
Each engagement follows the same shape: the problem, the approach, and what you are left holding at the end.
01
Application development
Problem — a process exists only in someone's head or a spreadsheet. Approach — model it, build it, hand it over. Result — a running system with the source code and documentation in your hands.
Available
02
Data and automation
Problem — the data exists but nobody can query it, and the same steps get repeated weekly. Approach — pipelines, storage, scheduled jobs. Result — reporting that updates itself.
Available
03
Applied AI integration
Problem — a judgement call made repeatedly by a person. Approach — model the decision, measure it against how it is made today. Result — a decision that is consistent and auditable.
Available
04
System architecture review
Problem — a build is about to start and the shape of it is not agreed. Approach — data model, boundaries, integration points, written down. Result — a specification your team or ours can build from.
Available
05
Research and prototyping
Problem — nobody knows yet whether the idea is technically viable. Approach — the smallest honest prototype. Result — evidence, and a recommendation to proceed or stop.
Available
Semiconductor tooling, autonomous simulation, space and financial infrastructure software are on our long-term roadmap and are not offered as services today. We list them under Work, marked as parked, rather than implying we can deliver them now.
04 — Work
Selected work and initiatives.
Status is stated on every item. Nothing here is presented as a finished commercial product before it is one.
01
ReturnIQ
Automates the product return lifecycle and decides the disposition of each item — repair, replace, refurbish, recycle, scrap or resell — using condition grading, cost prediction and fraud checks.
In development
02
THREAD
A campus platform for universities — the category occupied by Camu and similar student information systems. Where those record attendance and marks, THREAD is built around what students actually do with a course: learning paths, project work, and a record of how a subject was learned rather than only that it was passed.
In development
03
Gaze Glide
Computer control through an ordinary webcam using head movement, facial gesture and eye tracking, smoothed with a Kalman filter so the cursor is steady enough to use. Built for people with motor disabilities.
Prototype
04
Kalman Product Lab
Internal tooling used to build and test our own products — roughly 10,500 lines of TypeScript with a 48-check smoke suite that runs against the real client bundle rather than a mock. Not sold; it is how the other work gets made.
Complete
05
Interactive responsive web experience
A single-page scroll-driven site combining video, image and text, built to hold together from a 320px phone to a desktop display without a framework. Built as a working demonstration of front-end capability rather than as a product.
Complete
06
K-Drop
Peer-to-peer encrypted file transfer over WebRTC with a relay fallback. Installable as a progressive web app, English and Hindi, three dependencies total. Complete and working.
Complete
07
Chip Design Assistant · RTL Bug Finder · Verification Automation
Semiconductor engineering tooling. Requires domain access and EDA relationships we do not currently have.
Parked
08
Autonomous Multi-Agent Simulation
Agents representing people, businesses, transport and services, for city planning and supply chain testing. Years of work for a two-person team.
Concept
09
Space Software Platform · Financial Infrastructure Platform
Mission planning and telemetry; payment and risk infrastructure. Both require relationships and regulatory standing we do not have as an unfunded company.
Parked
05 — Areas of focus
Technology for industries that matter.
These are the sectors our work is aimed at and where our products are being built. We have no clients in them yet, so we call them areas of focus rather than industries we serve — the difference is deliberate and we will change the wording when it stops being true.
01
Manufacturing
Returns, warranty claims and recovery of value from defective stock.
02
Logistics & supply chain
Reverse flows, routing and warehouse decision-making.
03
Retail & e-commerce
Customer returns, disposition and fraud patterns at volume.
04
Education
Learning platforms and the resources students actually use.
05
Healthcare & accessibility
Assistive interfaces for people who cannot use standard input devices.
06
Financial services
Risk, fraud detection and payment infrastructure. Long-term, regulated.
07
Semiconductor
Design assistance and verification automation. Roadmap.
08
Aerospace & space
Mission planning, telemetry and ground software. Roadmap.
06 — Research
Researching what comes next.
Alongside client engineering, we work on problems that are not yet products. Two of these are active and have code behind them; the rest are areas we read, prototype in, and intend to work in as the company can afford to. Nothing here is published research and we do not claim otherwise.
01
State estimation and filtering
Kalman filtering applied to noisy real-time input — the basis of Gaze Glide's cursor stability, and of how we think about operational decisions generally.
Active
02
Decision modelling under uncertainty
Estimating cost and recovery for each possible outcome when the inputs are incomplete, then choosing a route that can be explained afterwards.
Active
03
Computer vision for interaction
Head, gesture and gaze tracking from ordinary webcam input, on hardware people already own.
Prototype
04
Autonomous multi-agent systems
Agents representing people, businesses and services, for simulating city planning, transport and supply chains.
Reading
05
Semiconductor design and verification
Where machine assistance genuinely helps in RTL work, and where it does not.
Reading
06
Simulation, robotics and space software
Mission planning, orbital visualisation and control software. Long-horizon interest, no current work.
Reading
07 — Approach
Built with engineering discipline.
Seven stages, the same every time. You should know what happens next at every point in the engagement.
01
Discover
Understand the problem and what it currently costs. We say plainly if it is not something we can build.
02
Define
Requirements, constraints and an explicit list of what is excluded. Fixed price, agreed date.
03
Architect
Data model, services and integration points designed before code is written.
04
Build
A weekly demo of software that runs, so direction is corrected early rather than at the end.
05
Verify
Reliability, security and performance tested against the constraints agreed at stage two.
06
Deploy
Delivered with source, documentation and a working deployment. It is yours.
07
Improve
An agreed support period after handover. Beyond that, ongoing work is optional and priced separately.
—
Throughout
One point of contact, and it is one of the two people writing the code.
08 — Status
Two engineers. One active product. No revenue yet.
You would find out anyway, so here it is first. What we can promise is that the people who take your call are the people who write the code, and that we will say when something is outside what we can do.
- Team
- Two engineers
- In development
- One active, one prototype
- Revenue
- None yet
- Legal status
- Pre-incorporation
Founded by Om Pilaji and Sarvajeet Bajikar, both studying computer science engineering at Woxsen University, Hyderabad. Om leads product direction and client engagement; Sarvajeet leads engineering and architecture.
09 — Vision
Building beyond the next release.
The plan is to grow Kalman from a technology consultancy into an engineering and research organisation building intelligent software systems across several industries — starting with one product that works, with real users paying for it, before the second one begins.
Consulting funds the products. Products earn the right to the next product. We would rather be a small company with something real running in production than a large-sounding one with a long list of announcements.
Built in India, designed for global problems. We are not trying to be an offshore development shop that happens to be cheap — we are trying to build software good enough that where it was written stops being the interesting part.
10 — Careers
We are hiring interns.
Kalman is two people. An intern here is not shadowing a team — you take a piece of real work, you own it, and it ships or it does not. You will be able to point at the specific thing you built. That is the trade we are offering, and it suits some people and not others.
We will not run an unpaid internship dressed up as experience. Stipend, hours and length are agreed with you in writing before you start, not after. Ask about them in your first message and we will answer with a figure rather than a range.
01
Software engineering — B.Tech CSE, IT, ECE
Backend services, frontend work and data pipelines on the returns platform. You will write code that runs against real return data, not a tutorial project. We expect one thing you have already built, with a link to the source.
02
Product and interface design — B.Design
The returns console is used by warehouse staff under time pressure, often on a low-end screen. The work is interaction design and information hierarchy for people who are not enjoying themselves. Portfolio matters more than software you know.
03
Spatial and warehouse systems — B.Arch, industrial design
Return intake, grading stations, storage and disposition are physical flows before they are software. We want someone who can read a floor plan and tell us where our model of a warehouse is wrong.
04
Marketing and communications
Positioning, written content and outbound to manufacturers and retailers. Our voice rule is plain and specific — no inflated language, no invented statistics. If you write that way already, say so with a sample.
05
Legal and compliance research — LLB, BA LLB
Research support on contracts, intellectual property assignment, data protection under the DPDP Act, and the terms we put in front of clients. This is a research role. Nothing produced here is legal advice, and all of it is reviewed by a qualified professional before it is used.
06
Operations and business analysis
Sit with the returns process at a real retailer, document what actually happens, and turn it into the rules the platform runs on. Suits commerce, management and economics students who like detail.
How to apply
Send one thing you built or wrote, the hardest problem you hit inside it, and what you want to be better at in a year. No CV template required, no covering letter. Write to om@kalmanconsultancy.com with the track name in the subject line.
We reply to every application within 24 hours, including the ones we turn down. Remote is fine. We are based in Hyderabad and the company is pre-incorporation, which we would rather you knew before you apply than after.
11 — Contact
Have a difficult problem?
Quote me in
Engagements are quoted per project, not per hour, so there is no price list to convert. Pick a currency and it travels with your enquiry — we will price and invoice in it. Currently set to INR.