I build systems that ship — from LLM pipelines to bare-metal RTOS.
Kamal A Sharma — Software Engineer — DevOps & AI Engineer. Full-stack, DevOps, and AI engineering with production mileage across five SaaS launches, a k3s cluster, and an embedded controller.
Profile
About the engineer
SoftwareEngineerwith2+yearsshippingproductionsoftwareacrossdatascience,backendengineering,DevOps,andembeddedsystems.Ilearnnewstacksbybuildingwiththem,notbyreadingaboutthemfirst:LLM-poweredpipelineswiththeOpenAIAPIandLangChain,PyTorchmodelsdeployedviaHelmonKubernetes,andanembeddedRTOSlearnedfromscratchforahands-onproject.Iwritedocumentationpreciseenoughthatsomeonewhowasn'tintheroomcanactonitwithoutaskingafollow-upquestion.
Looking for a role that uses this range — not one that filters for a single checklist skill.
- location
- "Frankfurt am Main, Germany"
- languages
- EnglishC1 · IELTSGermanB1 · Goethe-Institut, progressing to B2HindiNative
- status
- "open_to_work: true"
Career
Where I've worked
Software Engineer Associate (Full Stack & DevOps)
05.2022 — 03.2024Blackcoffer OPC Pvt. Ltd. · New Delhi, India
- Developed and deployed 5+ production-grade SaaS applications for global clients, improving client operational efficiency by up to 40%.
- Designed and implemented end-to-end CI/CD pipelines that reduced deployment time from days to under 2 hours.
- Containerized 10+ microservices using Docker, resulting in a 60% improvement in application scalability and deployment consistency.
- Provisioned and managed cloud infrastructure on AWS using Terraform, reducing infrastructure provisioning time by 70%.
- Automated monitoring and logging systems, decreasing average incident resolution time by 50%.
- Delivered 12+ major features on time across cross-functional teams, contributing directly to business growth and client retention.
Data Scientist Intern
12.2021 — 03.2022DevTown · Bangalore, India
- Built classification and regression models in Python (scikit-learn, pandas) end to end, from raw data to evaluated output.
- Designed reusable, reproducible preprocessing pipelines — feature engineering, normalisation, train/test splitting.
- Ranked top 20 of a 250-student cohort for consistent technical performance (per official letter of recommendation).
Selected Work
Things I've built
Intelligent Preprocessing Platform
Qualimind
A full-stack platform, shipped end to end, converting qualitative and mixed-type datasets into structured, ML-ready formats via a configurable R-Plumber pipeline with an LLM-powered step.
- Designed and shipped a full-stack platform end to end — concept to live deployment — with an LLM-powered step (OpenAI API, LangChain) and per-job metadata tracking.
- Built the React/TypeScript frontend and Express API with real-time job-status updates (Socket.IO); deployed a multi-environment stack on AWS EC2 (Terraform, Docker Compose, Nginx/TLS).
Distributed Edge-to-Cloud ML
Edge-Based Wildlife Detection
A distributed edge-to-cloud image classification pipeline running on a self-managed k3s cluster, with full observability across nodes.
- Built a distributed edge-to-cloud image classification pipeline (PyTorch) on a self-managed k3s cluster, persisting results to PostgreSQL and MinIO.
- Deployed via Helm on Kubernetes with Jenkins/GitHub Actions CI/CD and full Prometheus/Grafana observability across nodes.
Real-Time OS Washing Machine Controller
aRTS-RTOS-WMCTRL
A real-time, time-triggered controller built on an STM32, learning NuttX RTOS from scratch to implement a safety-focused finite-state machine.
- Learned NuttX RTOS from scratch to build a real-time, time-triggered controller on an STM32 NUCLEO-G431RB, implementing a safety-focused finite-state machine (door/vibration/sensor fault latches).
- Built a diagnostics dump reporting task-lateness and timing metrics to evaluate scheduling determinism under tight resource constraints.
MQTT Reliability Patterns
mqtt-server-sim
An MQTT pub/sub simulation demonstrating reliability patterns for availability-critical distributed sensor systems.
- Built an MQTT pub/sub simulation in C (libmosquitto) with QoS 1 delivery, Last-Will-and-Testament failure detection, and retained-message recovery.
- Containerised the Mosquitto broker with Docker Compose to demonstrate reliability patterns for availability-critical distributed systems.
Toolbox
Stack & capabilities
$ cat languages.txt
Certifications
AI Engineer for Developers — Associate
DataCamp · AIEDA0017792606213
Prompt engineering and LLM API integration for building AI-assisted application features and simple chatbots.
Data Engineer — Associate
DataCamp · DE0016442570088
Database design, SQL data management, cloud data storage/pipeline tools, data extraction, cleaning, and quality validation.
Docker Foundations Professional Certificate
Docker
Containerization fundamentals: images, containers, Dockerfiles, volumes, and networking.
Education
M.Sc. High Integrity Systems
Frankfurt University of Applied Sciences, Frankfurt am Main, Germany
04.2024 — 08.2027 (expected)
Relevant coursework: Distributed Systems, Data Analytics, Software Architecture, Secure Systems Engineering, Cloud Systems.
B.E. Computer Engineering
Gujarat Technological University, Surat, India
07.2019 — 05.2023
Relevant coursework: Database Management Systems, Software Engineering, Business Information Systems.