Rudrakshya Barman
Navigation
Rudrakshya Barman
Founder, Adytum & Principal Architect
Python · FastAPI · React · Flutter · Bare Metal GPU
I build software that
"I spent the first decade learning how to build software that solves real operational problems; the current decade is increasingly about learning how to turn that engineering ability into leverage, ownership and freedom."
Rudrakshya Barman
Founder, Adytum & Principal Architect
Python · FastAPI · React · Flutter · Bare Metal GPU
14+
Years building software
5+
Major product lines
2012
Builder journey begins
2026
AI + owned compute era
Strategic Blueprint
Structuring my engineering efforts around reliable cashflow engines, global & domestic product bets, and bare-metal compute ownership.
A real operating business I've built over 13+ years (~₹40 lakh annual revenue). Progati serves 100+ active schools (333 total base), while Adytum Health ERP (UTOPIA) provides steady, referral-driven hospital & lab workflows.
My distinct product growth strategies: Feedvoty as global SaaS (280+ signups), Bahify for India-specific GST/accounting and quick-commerce (Enjoy Mart, CuteClouds), and Voice AI agents (GOPRUKAI / NextAITel) for automated calling.
Transitioning from selling client engineering hours toward my owned bare-metal infrastructure, local GPU inference (Ollama, RTX 3090, L40S), credit-based AI features, and multiplying work output per hour.
Products & Companies
Platforms and products I have built around verified operational problems, sustainable cashflow, and scalable modern architectures.
School management ERP active across 100+ schools (333 total base). Built with constraint-solving timetabling, FastAPI ADMS attendance ingestion, Flutter mobile apps, and credit-based AI tools.
My longest-running healthcare domain solution for hospitals & diagnostics: OPD, IPD, laboratory/LIS, billing, prescriptions, patient records, and operational reporting.
India-focused platform I created combining sales, purchase, stock, GST, receivables/payables, accounting reports, multi-godown inventory, and store. Powering Enjoy Mart (quick-commerce), CuteClouds, and Mira Ayurvedic.
Customer feedback SaaS I built with FastAPI, PostgreSQL/RLS, and React. Reached 280+ sign-ups. Positioned as my primary candidate for international SaaS growth.
Adytuminfotech Softwares Pvt. Ltd. — my operating company for engineering, product development, and selective high-value technical consultancy. Targeting ₹10 Cr company valuation.
My AI calling platform for medical appointment booking and outbound demo scheduling: multilingual STT/TTS, silence handling, interruption detection, and caller context.
Strategic Eras
My evolution from hands-on software development to product ownership, bare-metal compute, and multiplying AI leverage.
PHP, MySQL, and jQuery foundation. I built software around real customer problems rather than tech for tech's sake. Early Adytum systems, Progati and UTOPIA.
Incorporated Adytuminfotech Softwares Pvt. Ltd. First office server (Dell T30 seed of self-hosting). My technical move toward Python, FastAPI, PostgreSQL and Flutter.
React + FastAPI + PostgreSQL/RLS default stack. Rebuilding mobile app experiences with Flutter. SaaS thinking and launching financial application software.
Progati scale (333 school base), healthcare stability, Feedvoty global experiment, Bahify formulation, and a deliberate product strategy.
Dedicated on-prem bare metal servers, local GPU inference (Ollama, RTX 3090, L40S, Blackwell), hybrid cloud architecture, voice AI agents, and credit-based AI in my products.
Moving from selling engineering hours toward products, owned compute, and applied AI that multiply the impact of every single hour of work.
Historical Record
My questions have moved upward: what deserves my time, what infrastructure I should own, which products can create true leverage, whether SaaS can escape local pricing constraints, how AI changes traditional ERP, and what kind of life all this technology is supposed to buy.
Sharpening my priority to Progati, Adytum Health ERP, and select high-value opportunities. Strategic market split: Bahify targeted for India (GST/accounting/local workflows), while Feedvoty remains my candidate for overseas SaaS growth. Expanding commercial reach from West Bengal to national level.
Rather than simply raising flat per-student monthly fees for price-sensitive Progati schools, I am leaning toward credit-based AI usage. The goal is to create tiny workflow advantages (voice marks entry, multilingual notices, generated reports/forms) that accumulate into undeniable operational value.
I explored an AI calling platform for medical appointment booking and outbound demo scheduling: multilingual conversations, STT/TTS, silence handling, interruption detection, caller context, and end-of-call detection. Evolving GOPRUKAI as R&D and NextAITel as a business layer.
AWS costs climbing above ₹42,000/month (including roughly ₹15,000 for Progati's c5.xlarge alone) triggered my infrastructure rethink. I decided cloud economics no longer made sense for these ERP workloads and designed a hybrid model: dedicated on-prem office servers, real-time cloud replication, nightly T30 backups, and AWS retained as DR.
My GPU research moved from curiosity to capacity planning: RTX 5060 Ti, RTX 3090, L40S, and RTX Pro 4000 Blackwell workstation designs. My core focus became assessing what inference capacity makes Progati, UTOPIA, and voice agents meaningfully more valuable.
Bahify developed into an India-focused platform combining sales, purchase, stock, GST, receivables/payables, accounting reports, and multi-godown inventory. Acquired my first three customers without marketing (Enjoy Mart, CuteClouds, Mira Ayurvedic). Enjoy Mart expanded into a full Blinkit-like quick commerce suite (Customer Android, admin web, delivery partner & reseller apps).
Attendance moved to a FastAPI ADMS ingestion service and PostgreSQL. Timetabling became its own FastAPI/Postgres/React system using constraint solving. I increasingly prefer separate applications connected around the core product instead of endlessly enlarging one monolith.
I structured early mornings around machine-learning study — often starting around 5 AM — working through classification, matrices, deep-learning foundations, and neural network history. I repeatedly connect theory back to products rather than learning it as an isolated academic exercise.
Adytum crossed its 13th anniversary — an unusually long operating history compared with the many experiments and products that come and go around it.
I tracked 333 schools in the overall list, 199 registered, 136 yearly customers, 59 monthly-app customers, and 101 schools using the app in some form. This sharpened a recurring tension: strong operational value versus an extremely price-sensitive school market.
I built Feedvoty with FastAPI, PostgreSQL/RLS, and React as a feedback SaaS. It reached 280+ sign-ups. I increasingly see Feedvoty as the product that could be pushed internationally.
React joined FastAPI and PostgreSQL as my stack for newer systems, SaaS products, and client software (including a financial loan application), while still maintaining mature PHP/MySQL products where rewriting everything would add little value.
I learned Flutter/Dart and rebuilt Progati's Parent and Teacher apps. The rewrite mattered because it was not cosmetic — the apps became significantly faster and more reliable than earlier versions. I also built Flutter apps for doctor booking, test booking, transport, and financial management.
Adytuminfotech Softwares Pvt. Ltd. was incorporated. My role increasingly became founder, principal builder, and technical lead. A Dell PowerEdge T30 began serving in my office for daily database backups — years later becoming the seed of a much larger self-hosting strategy. Python and FastAPI entered my stack.
The school ERP became a core Adytum product. Over the following decade, it expanded into fees, report cards, attendance, certificates, admit cards, timetables, notices, homework/classwork, transport, finance, ID cards, and parent/teacher workflows.
I began building software professionally using PHP, MySQL, and jQuery. Adytum's story began: building software around real operational problems rather than technology for technology's sake. I developed UTOPIA for hospital & diagnostic workflows (OPD, IPD, lab/LIS, billing, prescriptions, patient records, operational reporting).
Architectural Foundation
Pragmatic tech stack choices I've honed over 14 years: high-throughput APIs, decoupled microservices, and dedicated GPU compute.
Beyond Code
My engineering doesn't happen in a vacuum. Deep curiosity drives the underlying models behind my work across software, mathematics, and physics.
Exploring orthogonal matrices, Dirac matrices, and Minkowski space concepts that underpin modern physics and transformation mechanics.
Studying classification models, gradient mathematics, loss landscapes, and the pioneers who built early neural network architectures.
Pondering the nature of time travel, consciousness, dream states, and Vedic scientific assertions across ancient and modern frameworks.
Analyzing the societal consequences of automation, how software ownership creates independence, and building technology with purpose.
Occasional notes on software engineering, bare-metal AI infrastructure, product building, and creating leverage.