Skip to content

At Cefalo — open to select work

Mrityunjoy Das — Engineer.Mentor.Builder.

Backend and Data Engineer with 10 years of experience building scalable systems and robust data pipelines, now building the MCP tooling and agentic workflows on top of them. AI Coach and AI Task Force member at Cefalo, mentoring engineers on agentic development.

Dhaka, Bangladesh

Get in touchRésuméLinkedInGitHub
Portrait of Mrityunjoy Das

Engineering identity

  • Backend

    Backend services and the data models behind them

    REST APIs and microservices in Python — Django, Django REST Framework, FastAPI and Flask — with the data model designed before the endpoints. Multi-tenant architecture where the tenancy boundary is enforced in one place, single sign-on across services with Keycloak, Redis caching on measured hot paths, Celery and CeleryBeat for work that should never block a request, and Docker-based deployment. Most of the performance work has been query profiling and rewriting rather than adding capacity.

  • Data engineering

    Data pipelines that publish timeseries and forecasts

    200+ production pipelines behind Veyt's market intelligence, producing the timeseries, timeseries groups and forecast curves the platform sells across carbon, guarantees of origin, power purchase agreements and renewable fuels. An ingestion layer reaching dozens of external registries and APIs, Python ETL over sources with their own schedules and failure modes, Prefect orchestration, schema design for timeseries at volume, monitoring that surfaces a failed run before a customer does, multiprocessing work that took 50% off the slowest jobs, and query and index rewrites that took one core lookup from 99.4 ms to 0.193 ms.

  • AI

    MCP tools and agent workflows on live systems

    MCP servers that give agents authenticated, typed access to internal systems, so an agent reads the actual state instead of inferring it. Agentic workflows built on top of those tools for recurring engineering tasks, a production auto-fix agent hardened until most of its repairs land unattended, LLM API integration inside existing Python services, and coaching Cefalo teams through adopting all of it — including where the approach does not pay.

Experience

  1. Cefalo

    PresentDhaka

    Staff Software Engineer

    Present

    • AI Coach on Cefalo's AI pilot: mentored 12 colleagues on AI coding agents, agentic workflows, MCP tooling and prompting that holds up under real work.
    • Sit on Cefalo's AI Task Force, shaping how the company runs its AI transformation.
    • Built MCP tools that give agents first-class access to the systems behind Veyt's projects, rather than leaving them to guess.
    • Designed agentic workflows that take the repetitive work off the team — the recurring tasks that quietly consumed hours every week now run themselves.
    • Hardened a production auto-fix agent across ~50 automated repairs, closing the gaps in how it checked its own work and lifting the unattended fix rate towards 85%.
    • Set the technical direction for data engineering work across the Veyt account.
    • Built an HR agent for Cefalo at a company hackathon — an MCP-based RAG pipeline over the policy documents and the employee database, on LangChain, LangGraph, Pinecone, FastAPI and a local Ollama model.

    Senior Software Engineer · Data Engineer

    • Architected and scaled the ingestion layer that pulls from dozens of external registries and APIs, and feeds the analyst dashboards, forecasting and decision tooling built on top of it.
    • Built and run 200+ data pipelines producing the timeseries, timeseries groups and forecast data behind the platform's market intelligence.
    • Cut processing time on the slowest pipelines by 50%, by moving the work to multiprocessing and removing the code paths that made it serial in the first place.
    • Partitioned the large datasets so ETL runs in parallel, which brought down both the latency and the memory footprint of the high-volume executor jobs.
    • Rewrote the queries and indexes behind the production timeseries API: a core fragment lookup went from 1,141 rows read to 19, and from 99.4 ms to 0.193 ms — about 580 times faster — taking database compute from ~1.75 to ~0.91 CU-h per hour.
    • Put Redis in front of the heaviest reads and cut query response times by close to 60%.
    • Reworked the Dockerfile and brought build times down by 40%, which took a slow feedback loop out of every deploy.
    • Orchestrated every ETL workflow with Prefect, containerised the jobs, and shipped them to GCP Cloud Run through GitHub Actions.
    Veyt logo

    Veyt(Norway) — Independent Norwegian market intelligence provider for green certificates — carbon, guarantees of origin, power purchase agreements and renewable fuels. Over 300 firms across the Nordics and Europe price and benchmark against its data. Formerly Greenfact.

  2. Brain Station 23

    Dhaka

    Senior Software Engineer

    • Led 8 developers building a microservice, multi-tenant bus booking platform.
    • Found the queries that were dragging the system down and rewrote them, for a 50% overall performance gain.
    • Added a Redis caching layer that cut response times by close to 60% on the hottest paths.
    • Worked directly with product and design across teams to keep scope and architecture in step.
    Ferdia logo

    Ferdia(Norway) — Norwegian platform that lets bus companies and travel organisers run and connect their operations.

  3. ADN DigiNet Ltd.

    Dhaka

    Senior Software Engineer, Team Lead

    • Led a product development team of 5, from system design through feature planning with the business side.
    • Implemented single sign-on with Keycloak across a microservice architecture.
    • Built REST APIs in FastAPI and Flask, and containerised the services behind them.
    • Designed and shipped the billing module for a CRM product, plus reporting on Apache Superset.
    • Raised the team's baseline on code review, linting, and version control.

  4. InfoSapex Limited

    Dhaka

    Senior Software Engineer

    • Replaced the existing search with ElasticSearch and improved search accuracy by 42%.
    • Moved slow synchronous work onto Celery and CeleryBeat.
    • Shipped Django features used daily by some of Bangladesh's largest corporates.

    Software Engineer

    • Built and maintained a Document Management System, a hotel booking platform, and BRAC TBCP.
    • Added features to the booking platform's companion mobile app in Flutter.
    • Cut page load time by 20% by rewriting the worst of the front-end code.

    Software Engineer Intern

Products and systems

  • Data platform

    Veyt2022 – presentvia Cefalo

    200+pipelines in production

    The pipelines, APIs and agent tooling behind Veyt's published curves for green-energy markets.

    • 100%50%

      runtime, slowest jobs

    • 100%60%

      build time

    • 99.4 ms0.193 ms

      core lookup on the timeseries API

    Product
    Veyt sells market intelligence across every green-energy market it covers — carbon, guarantees of origin, power purchase agreements and renewable fuels. Traders and analysts act on the published curves the same day they land.
    What I did
    I architected the ingestion layer that reaches dozens of external registries and APIs, and I build and run the pipelines feeding all of it: Python ETL publishing timeseries, timeseries groups and forecast curves from upstream sources that each define a schedule and a format differently, orchestrated with Prefect and shipped to Cloud Run through GitHub Actions — including the sources that publish nothing and have to be scraped with BeautifulSoup and Playwright. I cut processing time on the slowest jobs by 50% with multiprocessing and by removing the code paths that forced them to run serially, and took 40% off build times by reworking the Docker layers. On the timeseries API I rewrote the queries and indexes — one core lookup went from reading 1,141 rows to 19, and from 99.4 ms to 0.193 ms — which took database compute from ~1.75 to ~0.91 CU-h an hour. I also build the REST APIs the platform exposes, and the MCP servers that give agents authenticated, typed access to the same systems.
    • Python
    • FastAPI
    • ETL
    • Prefect
    • Pandas
    • Playwright
    • Timeseries
    • MCP
    • PostgreSQL
    • Redis
    • Docker
    • GCP Cloud Run
    Case studyClose

  • Multi-tenant bus booking platform

    Ferdia2021 – 2022via Brain Station 23

    ~60%faster on the hottest paths

    Backend work across a Norwegian platform for bus operators — booking, fleet and scheduling.

    • 1x1.5x

      overall performance

    • 100%~40%

      response time, hottest paths

    Product
    A Norwegian platform that lets bus companies and travel organisers run and connect their operations — booking, fleet and scheduling, multi-tenant across every operator using it.
    What I did
    I worked across the backend — several of the services in the microservice split, with the tenancy boundary kept in a single layer instead of leaking into every one of them. Most of my time went on performance: I profiled and rewrote the queries carrying real load, then put Redis in front of what was left. Overall performance up 50%, and the hottest paths close to 60% faster.
    • Python
    • Microservices
    • Redis
    • PostgreSQL
    Case studyClose

  • Roboket — CRM, ads manager and reporting suite

    2021via ADN DigiNet Ltd.

    A distributed product suite behind one sign-on: CRM, ads manager, URL shortener and reporting.

    Product
    Roboket is ADN Telecom's marketing product suite — a CRM, a Facebook ads manager, a URL shortener and a reporting module, each its own service, all behind a single sign-on.
    What I did
    I led the team of five and built the backend services in FastAPI and Flask, with SQLAlchemy over the data layer and Celery for the work that had no business blocking a request. I put Keycloak in front of the whole suite so one account opens every module, wired the ads manager to the Facebook Marketing API, shipped the CRM billing module, and stood the reporting up on Apache Superset.
    • Python
    • FastAPI
    • SQLAlchemy
    • Celery
    • Keycloak
    • Apache Superset
    • MongoDB
    • Docker
    Case studyClose

  • Document management, hotel booking and BRAC TBCP

    2016 – 2020via InfoSapex Limited

    +42%search accuracy

    Three Django products for Bangladesh's largest corporates, and the Flutter app that goes with the booking one.

    • 100%80%

      page load time

    Product
    Django products used daily by some of Bangladesh's largest corporates — a document management system, BRAC's TB Control Programme, and a hotel booking platform with a companion mobile app.
    What I did
    I built these products end to end, and shipped features on the Flutter companion app alongside them. I replaced the search backend with ElasticSearch, where most of the work went into analysers and field weighting rather than the swap itself, for a 42% accuracy gain. I moved the long-running jobs onto Celery and CeleryBeat so requests stopped waiting on work nobody was watching, and rewrote the worst of the front-end code where it was the real cause of a slow page.
    • Django
    • Flutter
    • ElasticSearch
    • Celery
    Case studyClose

Recommendations

Worked together on the same team at BS23. Now we work in different teams at Cefalo, yet we still meet and discuss quite a bit, sharing ideas, challenges, things we are learning, yada yada.

Technically, he is my senior in terms of experience, but I have always found him to be a very good buddy. He talks less, thinks more, and delivers even more.

He has strong expertise in Python backend development, cloud technologies, and data engineering, and has contributed a lot in senior engineering roles. Polite and very helpful, he also brings a great mentoring spirit. It would have been great if we could work together on the same team again :)

See moreSee less

Abdullah Al Masud Tushar

Product-Minded Engineer Helping European Teams Build Reliable, High-Impact Systems | Fullstack & Data Engineering

Worked with me but on different teams

Mrityunjoy stepped into our team with massive shoes to fill, taking over the role of the most senior resource in the team. Some had their doubts initially that this was even a role that could be properly filled by someone else, but Mrityunjoy took it upon himself to ensure that the doubters could not have been more wrong!

He has earned my utmost respect, being a person who will always step up to the plate when things get tough, and will see things through to the end. Mrityunjoy does not back down from a difficult situation, and rather takes charge and displays great leadership!

He is a rock solid person with an unmatched can-do attitude, and has integrity that earns respect. Even faced with critical, time-sensitive issues, Mrityunjoy keeps calm, collected, and confident until the task is seen all the way through.

You've undoubtedly earned my respect, both for who you are as a person, as well as your knowledge in what you do in your work. It has been an absolute pleasure working with you, and I would say "good luck" moving forward, but you don't need luck!

See moreSee less

Erik Ormevik

Product Owner & Frontend Developer at Ferdia

Worked with me at Ferdia

Skills

  • Backend & Languages

    • Python
    • Django
    • FastAPI
    • Flask

  • Databases

    • PostgreSQL
    • MySQL
    • MongoDB
    • Redis
    • Pinecone

  • Data Engineering

    • ETL
    • Prefect
    • Pandas
    • Celery
    • Elasticsearch
    • Apache Superset

  • Agentic & AI

    • Claude Code
    • MCP
    • Agent Architecture
    • Agentic Workflows
    • LLM Integration
    • Prompt Engineering
    • RAG
    • LangChain
    • LangGraph
    • Ollama

  • Infrastructure

    • Docker
    • AWS
    • GCP
    • Microservices
    • Keycloak
    • GitHub Actions
    • CI/CD
    • Git

Certifications

Writing

Education

  • B.Sc. in Computer Science and Engineering · National University of Bangladesh

    Apr 2016Bangladesh

    Under: Institute of Science and Technology

Get in touch

Let's talk about your project.

I'm always happy to talk about backend architecture, data pipelines, agentic systems, or whatever is quietly costing your team its afternoons.

mjoyshuvo@gmail.comBook a 30-min call
in/mrityunjoy-das@mjoyshuvoDhaka, Bangladesh