Pune, India

Pranav Gujar — Software Engineer, backend systems and applied AI, 2 years 9 months in production, Pune, India

Software Engineer — Backend systems & applied AI

I build production Python systems — APIs, data pipelines and AI workflows — and take them from initial requirements through to deployment and ongoing support. Currently at Blueport Solutions, where I own the delivery of a multi-tenant SaaS platform end to end.

See the work Resume

Backend: Python / FastAPI / Flask / Django REST
Data: PostgreSQL / Redis / MongoDB / PgBouncer
Applied AI: LangGraph / RAG / Hugging Face / ONNX
Infra: Docker / AWS / Temporal / RabbitMQ / CI/CD

Python
FastAPI
LangGraph
PostgreSQL
Redis
Temporal
RabbitMQ
Docker
AWS
ONNX
Pytest
GitHub Actions
Python
FastAPI
LangGraph
PostgreSQL
Redis
Temporal
RabbitMQ
Docker
AWS
ONNX
Pytest
GitHub Actions

2Y 9M: Experience, since Feb 2024
10+: Client applications delivered end to end
5+: Applications I run in production today
3+: Systems I designed, built and operate
2: AWS certifications

01Selected work

Projects

Three systems I designed, built and operate. In each, the defining problem was a constraint to work within rather than a feature to add.

01 / 2026

Evaluation platform

EIGENN

A platform built for a single domain: competitive-exam preparation. Coaching institutes enrol and grant access to their students; a student submits a handwritten answer sheet and receives an evaluation in one step, with a human reviewer approving every AI result before the student sees it.

FastAPILangGraphTemporalONNXReact

Built a multi-tenant platform serving two distinct user groups — coaching institutes who enrol and administer their own students, and the UPSC, state PSC, CA and CS candidates those students are.
Engineered a 17-node LangGraph pipeline that reads a scanned answer sheet, reviews its own intermediate output, declares every failure state, and resumes from any node rather than restarting — instrumented end to end with Langfuse.
Built retrieval over 120K+ chunks using BGE-M3 with cross-encoder reranking, quantised to ONNX INT8 to run on CPU within 300MB — removing GPU cost from the platform entirely.
Architected the backend as 10+ containerised services on a private IPv6 mesh behind a React and TypeScript front end, with long-running evaluations made durable on Temporal.
Implemented idempotent Razorpay payments over an append-only ledger so retries are safe, supported by 150+ tests including contract tests that detect changes in AI output structure.

02 / 2025–present

Market screener

AlgoPy

Runs 16 independent screening strategies across 8,100 NSE and BSE companies each night, publishes each as a pre-computed snapshot, and states the measured backtest result on the page — including where it shows no edge.

Django 5PlaywrightGitHub ActionsDockerPWA

Publishes its own negative results. Each strategy states what its backtest measured — t-statistics, profit factors, multiple-testing corrections — including the screens that demonstrate no edge.
Serves 8,100 companies in under 0.3s on a 512MB host by committing each night’s data into the deployment, so no external call occurs while a reader waits.
Reduced serve time 200s → 0.7s, memory 62MB → 27MB per worker, and upstream requests 13,000 → 3,000 per night. Each figure is a measurement taken before and after.
Covered the platform with 2,341 automated tests, 147 of which drive a real browser, and operate the whole system at ₹0/month within free-tier limits.

03 / 2025

Resume engine

AgentCV

Reads a job posting and assembles a tailored, ATS-ready resume and cover letter from achievements I have already written, then assesses the fit requirement by requirement and explains the result.

Flaskpython-docxWeasyPrintPlaywrightDocker

Designed, built and shipped independently in five weeks — 318 commits, 59 routes across 10 screens, 5 document templates. Live, and in use for every application I submit.
Built a matching engine that explains its own output. Each requirement in a posting is classified as evidenced, mentioned, available but unused, or a genuine gap, so every score identifies the sentence behind it.
Developed a skill vocabulary of 211 canonical terms and 614 surface variants, together with detection for technologies outside it, so no requirement is silently discarded.
Wrote 46 test suites, 9 of which measure rendered output in a real browser across 10 device widths — identifying overlapping text and sub-44px tap targets that static analysis cannot.

02Profile

About me

Based: Pune, India
Open to: Full-time roles
Relocation: Anywhere in India
Notice period: 60 days
Work mode: Onsite, hybrid or remote
Experience: 2 years 9 months

I am a software engineer based in Pune, working on backend systems and applied AI. Most of my work is in Python: HTTP APIs, data pipelines, background workers, and the retrieval and evaluation layers around language models.

I take responsibility for the full stack of a system rather than a single layer: the database schema, the API, the retrieval and evaluation around a model, and the work of keeping it fast and correct once it carries real traffic. Recent projects include a 17-node LangGraph evaluation pipeline, retrieval over 120,000+ chunks running entirely on CPU, and an OCR pipeline that migrated five years of legacy records into PostgreSQL without the loss of a single row.

I also operate what I build — deployments, certificates, logging, and the on-call response when something fails at an inconvenient hour. Being accountable for a system in production is what taught me which shortcuts are affordable and which are not.

03Experience

Where I have worked

Aug 2025 — Present

Blueport Solutions

Pune, India

Software Engineer

Own the delivery of a white-label SaaS platform that serves several tender-management firms from one codebase, with each client’s data kept separate and access set by role.
Cut the average API response time by 70% with Redis caching, indexing, load balancing and faster queries.
Built an OCR pipeline that moved five years of old spreadsheets and PDFs into PostgreSQL without losing a row.
Built a RAG pipeline that reads tender listings and shows each client only the ones in their own field.
Run the VPS setup on Coolify — SSL/TLS, versioned releases, logging in one place. Automating the release steps took about 20 minutes off every deploy.

The product I work on

Sutram

Tender discovery and bid management for government contracts — identifying matching tenders, reading their documents, and carrying the work through to payment.

View details

Feb 2024 — Jul 2025

Beginning Solutions

Pune, India

Full Stack Developer

Delivered 6+ client applications end to end — scope, design, build, deployment and maintenance.
Built REST APIs in Django REST Framework with JWT auth, wired up to Razorpay, OpenAI and Google APIs.
Managed hosting for 10+ live applications on Hostinger VPS and AWS EC2; faster queries cut page load times by 30%.
Ran full Google Play release cycles — signing, staged rollouts, pre-release testing, and crash and ANR monitoring.
Migrated a production database off a legacy schema with Python written for the job, keeping every existing user record intact.

The product I worked on

DMFU

Daily Mess For U

Two Android applications addressing the same problem: locating an affordable daily meal, and administering the mess that serves it.

View details

04Stack

What I work with

Languages: PythonJavaScriptSQL
Backend: FastAPIFlaskDjangoDjango REST FrameworkREST API designRazorpay integrations
Applied AI: LangGraphRAG pipelinesHugging FaceChromaDBLangfuseONNX INT8
Frontend: ReactTypeScriptVanilla JSTailwind CSSPWA / service workers
Data: PostgreSQLMySQLMongoDBRedisPgBouncerSchema migrations
Distributed: RabbitMQTemporalBackground workersAsync queues
Cloud & DevOps: AWS (EC2, S3, RDS, VPC, Route 53, IAM)DockerGitHub ActionsLinux
Testing: PytestPlaywrightContract testsGolden testsBacktests
Web & distribution: Technical SEOStructured data (schema.org)Google Search ConsoleGoogle Tag ManagerGoogle Play ConsoleCore Web Vitals
Operations: VPSCoolifyApacheSSL/TLSLoggingMonitoringReleases

Education

MBA — Business Analytics
2022 — 2024
Pune University
CGPA 8.02
BSc — Computer Science
2019 — 2022
Pune University
CGPA 7.37

Certifications

05Contact

Get in touch

I am seeking a full-time backend engineering or applied AI role and am open to relocation anywhere in India. Email is the quickest way to reach me.

Email: pranavgujar2901@gmail.com
Phone: +91 91466 24414
GitHub: pranavg2901
LinkedIn: in/pranavpgujar

Based: Pune, India
Relocation: Anywhere in India
Notice period: 60 days
Work mode: Onsite, hybrid or remote

Send a message