Skip to content
Akshith Mysa

Magdeburg, Germany · ——:—— CEST

Akshith Mysa

Systems engineer working two layers at once — production platforms, and the silicon they run on.

Building
Co-founder & CTO, Klinn AI — client platforms and SEO sites in production since 2025
Studying
MSc Advanced Semiconductor Nanotechnologies, OVGU Magdeburg
Published
Wiley, 2025 — autonomous ornithopter navigation (ch. 20)

01 / Work

Selected systems, in detail

Two full-stack client platforms, SEO landings for consultancies and pharma, Genau for German, an earlier RAG experiment, and the navigation work that became a Wiley chapter.

01Client · Full-stack2025–26

Edmissions World CRM

Klinn AI's production staff CRM, live at crm.edmissionsworld.com. I owned architecture, data model and delivery.

Read case study →
Problem
An education consultancy ran its admissions floor across spreadsheets, WhatsApp and memory. Five jobs — calling, qualification, documents, walk-ins, management — lived in five places.
Built
A private, role-based CRM: shared lead pipeline, telecaller qualification, counselor documents, reception walk-ins, WhatsApp templates, and an overnight job that reopens no-contact leads.
Outcome
In production at crm.edmissionsworld.com. Admin, managers, telecallers, counselors and reception work one pipeline instead of three spreadsheets.
Next.js 15TypeScriptFirebaseCloud FunctionsLiveKit

02Client · Full-stack2026

Metaflow

A multi-tenant business platform for a client — website, admin, payments, industry workflows, publish.

Visit site →
Problem
The client needed non-technical customers to stand up a professional site without a developer on every change, and without one tenant ever seeing another tenant's data.
Built
A SaaS control centre: pick a template, generate a category-shaped site, edit it on a Craft.js canvas, pay on Stripe, publish to a subdomain or custom domain. Postgres is tenant-scoped; an AI assistant drafts copy server-side.
Outcome
Live at metaflow.in. The product is the builder — a full-stack platform, not a one-off brochure.
Next.jsPostgresStripeCraft.jsGCP

03Hobby · in progress2026

Genau

A native iOS German trainer I am building to learn the language and to practise speaking it.

View source →
Problem
Streak apps drill words I will never say at the Bürgeramt, and they go quiet the moment I actually need to talk. I wanted something I would open on the tram.
Built
A SwiftUI app on SwiftData: noun gender drills (der / die / das), verb flashcards, a dictionary I can seed, and spoken feedback so I hear the form. Conversation practice is the part I am adding next — the point of the project, not a badge.
Outcome
Ongoing. I use it to study German and to get the sentences into my mouth, not to keep a streak alive.
SwiftUISwiftDataSpeechiOS

04Hobby2024

PDF Buddy

An earlier RAG experiment — upload a document, ask it questions, keep the thread.

View source →
Problem
Reading a long PDF still meant scrolling, grepping, and losing the thread. Chatbots that 'read' the file once had no memory.
Built
A Next.js app where you drop a document, LangChain chunks it into Pinecone, and Gemini (or OpenAI) answers with chat memory. Clerk for auth, Firebase for files.
Outcome
A working companion from 2024: summaries, multi-turn questions, and the document still on screen. I have since moved on.
Next.jsLangChainGeminiPineconeClerk

05ResearchTiHAN

Autonomous Ornithopter Navigation

Research intern at TiHAN — IIT Hyderabad, India's autonomous navigation hub.

Read the paper →
Problem
Flapping-wing UAVs oscillate along every axis by design, so the pose estimation that works on quadrotors degrades badly — and indoors there is no GPS to fall back on.
Built
A navigation and control stack for indoor and outdoor flight: UWB ranging for absolute position, ORB-SLAM3 with a RealSense D435i, ROS for control, tuned against the airframe's own flapping period.
Outcome
Published as chapter 20 in Artificial Intelligence Applications in Aeronautical and Aerospace Engineering (Wiley, 2025).
ROSORB-SLAM3UWBPythonFlight testing

Also shipped: AI LMS (adaptive learning platform), DroNavNet (drone navigation nets), and Siamese face ID (one-shot embeddings). Source on GitHub.

02 / Research

What I want to do next

01

Neuromorphic & in-memory compute

Devices that do the arithmetic where the data already sits. I have spent two years on the software side of inference cost; this is the other end of the same problem.

02

Nanoscale device characterisation

Fabrication and measurement of advanced semiconductor devices — the core of the OVGU programme, and the part I want to be genuinely good at.

03

Autonomy under constraint

Navigation and control when the sensors are poor and the platform is unhelpful. Where my published work sits, and still the problem I find hardest to put down.

Peer-reviewed · Wiley 2025

Navigation of Unconventional Drones — Autonomous Ornithopter

S. Syam Narayanan, P. Rajalakshmi, Y. Gangurde, A. Mysa, S. Movidi

Read on DOI ↗

03 / Path

Where I have been

  1. Current

    MSc Advanced Semiconductor Nanotechnologies — OVGU Magdeburg

    Device physics, fabrication and characterisation. The layer underneath everything I had built until then.

  2. 2025 — now

    Co-founder & CTO — Klinn AI

    Architecture and delivery for production platforms: a staff CRM, a multi-tenant business builder, and SEO landings for clients who needed to rank. Software that has to work on a Tuesday, not just in a deck.

  3. 2024 — 25

    Associate AI Engineer — OSCOWL ai

    Applied AI in a deep-tech lab working across speech models, UAVs and custom silicon. Research that had to survive contact with a product.

  4. TiHAN

    Research Intern — TiHAN — IIT Hyderabad

    Bio-inspired flight: autonomous ornithopters, GPS-denied navigation, UWB and SLAM. Led to the Wiley chapter.

  5. 2022

    Research Intern, Dept. of ECE — NIT Raipur

    Signal processing and deep learning on EEG time-series for brain–computer interfaces — preprocessing and feature extraction in Python.

  6. 2020 — 24

    B.Tech. Electronics & Communication Engineering — KL University

    Devices, signals, and the first drones. The undergraduate that got me into the hardware underneath the software.

Portrait of Akshith Mysa

Most of the last two years went into shipping. Klinn AI is the company I co-founded to build production software for operations that cannot afford theatre — full-stack platforms and the SEO sites that feed them. Before that I was in Hyderabad's deep-tech pocket: bio-inspired drones at TiHAN, then applied AI at OSCOWL. The master's at OVGU is the layer I kept skipping — models and products are downstream of devices.

05 / Contact

If the problem is still awkward, write.

Open to doctoral positions, engineering roles in the EU, and conversations about hard systems. Magdeburg afternoons, Hyderabad evenings — I answer both.

© 2026 Akshith Mysa

Magdeburg · ——:——