10,000+ tags a second on a 150,000-tag platform: Python ingestion into MongoDB and Azure IoT Hub for one of India's largest steel manufacturers, owned end to end as DRI. Designed the edge-to-cloud architecture and wrote the ingestion.
Saptashwa Bandyopadhyay
Backend and distributed-systems engineer · 9+ years · led a team of ~8 · platforms for 10,000+ employees
- Now
- Manager, Architecture & DRI Lead at PwC India, since April 2026
- Before
- Senior Associate, team lead and platform owner at PwC India, 2022–2026
- Based in
- Kolkata, India
The PDF has the same content as this page, in one plain column that applicant tracking systems can read. 2 pages · updated Nov 2026.
Headline work
15× faster semantic search (30+ s to under 2 s), adopted by 90% of 10,000+ employees across internal tools that were searched separately before.
30,000+ user accounts onboarded through one identity and access module reused by 3 platforms; designed its OAuth token management, session lifecycle and per-application session limits.
Roles
newest firstManager, Architecture & DRI Lead
Industrial IoT data platform (steel manufacturer)
DRI and architecture owner, hands-on in Python.
- Cut end-to-end latency from 8+ s to under 1.5 s (5×) across the Kepware, Ignition and Azure IoT Hub pipeline.
- Designed dual writes to MongoDB and Azure IoT Hub so data survives the failure of either sink; owns MS SQL schema design and query tuning.
- On-call escalation point; wrote the design documents, runbooks and incident procedures across 3 integrated systems.
Pre-sales and solution architecture
Leads pre-sales and solution architecture, about 30% of the current role.
- Won 3 of 3 bids since leading pre-sales in September 2025: proposals, solution designs, estimates and client demos.
Senior Associate, Team Lead & Platform Owner
AI semantic search and insight platform (firm-wide)
DRI and platform owner, leading a team of ~8 developers.
- Built semantic search on Weaviate and PostgreSQL pgvector, evaluated relevance with a labelled query set, and tuned chunking and hybrid search.
- Self-hosted GitLab CE and Runners and built CI/CD from scratch, after finding the department's instance could not reach the deploy VMs.
- Worked directly with 2 Partners and 3 Directors on requirements and priorities.
Identity and access module (3 platforms)
Main designer of the module that connects each application to the firm's identity platform.
- Worked closely on implementing RBAC; implemented SSO federation for client organisations and MFA.
Enterprise SaaS (contract management, healthcare)
Backend engineer and technical lead for code and security reviews, for 5+ enterprise clients.
- Designed multi-tenant PostgreSQL with a schema per client, and moved long-running work onto RabbitMQ and ActiveMQ.
- Resolved 500+ issues as technical lead for code and security reviews; led hardening that cut vulnerabilities by 70%.
US banking (two banks)
Two separate US banks: hands-on builder at one, DRI for a platform migration at the other.
- Built a Python pipeline converting SWIFT and Excel payment formats at 99.9% accuracy across 500+ fields per file.
- Automated validation from 2–3 developers × 2 days to under 10 minutes.
- Led the Jenkins to GitLab CI/CD migration: builds 400% faster for 20+ engineers; set up and managed the Kafka pipeline.
AWS serverless app (steel pipes manufacturer)
Worked with the client's managers and shop floor on what to measure, then designed and built the app.
- Built Python Lambda services taking machine events from AWS IoT Core into DynamoDB, with API Gateway endpoints for real-time access.
Earlier roles: Associate, PwC India (2018–2022) · Software Systems Analyst, Open Source Software Solutions (2017–2018)
Associate
Backend and image-processing projects for enterprise clients.
- Improved licence-plate OCR accuracy by 40% with a custom-trained KNN model for a smart parking system.
- Built Node.js microservices for a contract lifecycle SaaS serving 5+ clients, with digital signature, DMS and BPM integrations.
- Built a code review tool on SonarQube and Pylint, and document and barcode processing in Python and Flask.
Software Systems Analyst
CRM data sync platform (HubSpot and Microsoft Dynamics)
- Designed and built a bidirectional HubSpot and Microsoft Dynamics sync engine on a 10-minute cycle for several enterprise clients, cutting sync time by 65%.