Architecture & Platform Design
Design cloud-native SaaS platforms, REST and gRPC APIs, microservices and distributed systems, with a focus on scalable architecture, security and reliability.
Dinesh Mandal
Exploring what's next. Engineering what lasts.
A technology enthusiast and Principal Software Engineer with 12 years of experience building cloud-native SaaS platforms, APIs and distributed systems across healthcare, advertising and enterprise technology domains. I've led the architecture and delivery of platforms processing 10B+ events daily at 99.99% availability, and built AI-powered applications using LLMs, RAG and AI agents. Working primarily with Go and Python, I stay hands-on, writing the code behind the systems I architect.
- Experience
- 12 years
- Currently
- Winter Inc ↗
- Previously
- Microsoft · Xandr (AppNexus) · AT&T
What I bring
From architectural vision to production reality.
Taking ideas from concept and MVP (Minimum Viable Product) to production‑scale systems through hands-on engineering and technical leadership.
Engineering & Applied AI
Build backend services and high-throughput data pipelines using Go and Python, alongside AI-powered applications leveraging LLMs, RAG and AI agents.
Technical Leadership & Delivery
Lead technical direction, drive architecture and code reviews, establish engineering practices, and mentor engineers while collaborating across Product, Operations and Engineering.
Experience
Career journey
Most recent first.
Sep 2025 — PresentChennai, India · Remote
Principal Software Engineer
Winter Inc.
- Apr 2026 — PresentWinter — continued in role after Winter acquired PeerIslands.
- Sep 2025 — Mar 2026PeerIslands — joined as Principal Software Engineer.
- Led architecture for cloud-native REST/gRPC API platforms in Go and Python, improving service availability to 99.99% and standardizing delivery practices across 3 teams for secure healthcare SaaS integrations.
- Redesigned RCM user management with hierarchical RBAC and module-level authorization — 60% less access-management effort and 80% fewer privileged-access incidents, with stronger HIPAA-aligned security controls.
- Built an AI-driven claims cohorting platform on historical RCM data to surface denial risk and resolution patterns — 25%+ higher representative productivity and ~20% faster average claim resolution.
- Remediated findings from a Thoropass-conducted penetration test across web, API and LLM surfaces, implementing security hardening that supports HIPAA and SOC 2 compliance objectives — with no Critical or High findings open following remediation.
- Owned API architecture and delivery with Product, Operations and Engineering, and mentored distributed engineers to improve production readiness and release quality.
Jun 2022 — Aug 2025Bengaluru, India · Remote
Software Engineer II
Microsoft
- Led a distributed diagnostics platform in Go and Python processing 10B+ log events per day, reducing real-time aggregation and query latency by ~30%.
- Designed secure Go backend services with OAuth 2.0, JWT, RBAC and Azure AD, sustaining 99.99% availability and cutting access-related support effort by ~30%.
- Developed production AI agents with Azure OpenAI, RAG, Azure Cosmos DB vector storage and prompt engineering that translate natural-language requests into optimized SQL/NoSQL queries through scalable backend APIs.
- Designed production LLM integrations for document understanding, retrieval and prompt optimization — enabling natural-language reporting over enterprise data and reducing manual analysis effort by ~25%.
- Led cross-functional AI initiatives across infrastructure, product and data science, reducing issue investigation and resolution time by 60%.
Jun 2021 — May 2022Bengaluru, India · Remote
Software Engineer
Xandr
Formerly AppNexus · Acquired by Microsoft
- Built high-throughput streaming services in Go, Python and Kafka processing 10B+ real-time auction events daily for globally distributed ad delivery.
- Designed Go microservices and APIs with Redis caching for millisecond-latency delivery, plus analytical schemas for near real-time reporting.
May 2020 — Jun 2021Bengaluru, India · Remote
Software Engineer
AT&T
- Built distributed Go and Python services and ETL pipelines processing high-volume advertising events and user-interaction data.
- Tuned SQL and MongoDB workloads and engineered parallel Go processing for large-scale cost and conversion metrics.
Sep 2014 — May 2020Bengaluru, India
Earlier roles
- May 2019 — May 2020Tata Consultancy ServicesIT Analyst
- Sep 2017 — Apr 2019TeamLease DigitalAssociate Consultant
- Sep 2014 — May 2017Capgemini EngineeringAssociate Consultant
Core expertise: Enterprise data warehousing and BI, ETL/ELT pipeline design, dimensional and star‑schema modelling, analytical SQL optimization, and governed reporting across multi‑million‑record customer and operational datasets, including supply‑chain analytics.
Education & certifications
DegreeMaster of Computer Applications (MCA)Vellore Institute of Technology2011 – 2014
DegreeBachelor of Computer Applications (BCA)Birla Institute of Technology, Mesra2008 – 2011
CertificationMongoDB Certified Developer AssociateMongoDB Inc.December 2025
Additional credentials
AWS Certified Solutions Architect – AssociateAmazon Web Services · Earned 2015
Selected engineering work
Case studies
A closer look at the engineering decisions, systems and measurable outcomes behind my work.
Security engineering across web, API and LLM surfaces
Strengthened a healthcare SaaS platform's security posture by addressing findings from a Thoropass-conducted penetration test, implementing targeted security hardening across web, API and LLM surfaces to support HIPAA and SOC 2 compliance objectives.
What I built
- Enforced server-side function- and object-level authorization across API routes, with responses scoped to role-appropriate fields.
- Added LLM guardrails against prompt injection by protecting system prompts and sanitizing persisted AI content.
- Hardened server-side handling of user-defined transformations and data processing.
- Strengthened transport and token security through HTTPS-only APIs, short-lived JWTs and least-privilege token audiences.
- Improved browser and file security through Content Security Policy, formula-safe CSV exports and authenticated signed storage URLs.
Tech stack
- Go
- Azure
- Auth0
- Docker
- Kubernetes
Security engineering
- JWT / OIDC
- TLS
- Content Security Policy
- OWASP API Top 10
- OWASP LLM Top 10
Security assessment
Thoropass-conducted penetration testing across web, API and LLM attack surfaces.
Remediation outcome
No Critical or High findings open following remediation.
Claims cohorting: deterministic scoring, AI-assisted explanations
Revenue-cycle teams needed to identify groups of open claims with elevated denial risk and unpaid balances, supported by actionable evidence for daily resolution workflows.
Process flow
- Feature extraction
- Metrics
- Cohort discovery
- Live validation
- Evidence
- AI summaries
What I built
- Developed a daily, per-client Databricks pipeline over MongoDB for feature extraction, metric calculation and cohort discovery.
- Implemented incremental processing with checkpoints, a two-day overlap and MongoDB
$mergeto process only changed claims. - Built multidimensional aggregation across payer, claim age, balance, CPT and denial themes, with five deterministic strategies ranking candidates using percentile-based metrics.
- Revalidated candidates against live open claims and linked them to claim-level audit evidence, using stable fingerprints to maintain consistent pattern identities.
- Integrated Azure OpenAI exclusively for pattern naming and daily summaries using group-level totals. Scoring remained deterministic and no PHI was sent to the model.
Tech stack
- Go
- Python
- Databricks
- MongoDB
- Azure OpenAI
- Azure
- Docker
- Kubernetes
Engineering approach
- Incremental processing
- Deterministic, rule-based scoring
- Multidimensional aggregation
- Cohort discovery
25%+improvement in claims representative productivity
~20%reduction in average claim resolution time
Hierarchical RBAC for RCM user management
RCM user management required finer-grained authorization across organizational roles and application modules, while maintaining strong HIPAA-aligned security controls.
What I built
- Redesigned the RCM permission model around hierarchical roles and module-level authorization.
- Implemented fine-grained access policies across RCM modules to align permissions with user responsibilities.
- Strengthened privileged-access governance through least-privilege controls and improved authorization boundaries.
Tech stack
- Go
- Azure
- Docker
- Kubernetes
Security architecture
- Hierarchical RBAC
- Module-level authorization
- Least-privilege access controls
- HIPAA-aligned security controls
60%reduction in access-management effort
80%reduction in privileged-access incidents
Real-time diagnostics for distributed systems
Operational analytics over billions of daily diagnostic log events required scalable aggregation and low-latency querying.
What I built
- Led the development of a distributed diagnostics platform using Go and Python.
- Designed high-volume ingestion and aggregation pipelines for operational analytics.
- Optimized real-time aggregation and query execution paths to reduce end-to-end latency.
Tech stack
- Go
- Python
- Kafka
- Azure
- Docker
- Kubernetes
Architecture
- Distributed event processing
- Real-time aggregation
- Low-latency querying
10B+diagnostic log events processed per day
~30%reduction in aggregation and query latency
NL-to-SQL AI agents over enterprise data
Reporting over enterprise data depended on manual analysis and hand-written queries.
What I built
- Developed production AI agents that translate natural-language requests into optimized SQL/NoSQL queries.
- RAG with Azure Cosmos DB vector storage; prompt engineering and optimization.
- LLM integrations for document understanding and retrieval, served through scalable backend APIs.
Tech stack
- Python
- Azure OpenAI
- Azure Cosmos DB
- Azure
- Docker
- Kubernetes
AI engineering
- RAG
- Vector search
- Prompt engineering
- NL-to-SQL / NoSQL
~25%less manual analysis effort
Technical expertise
Skills
The technologies I use to design, build and scale production systems.
Languages
- Go
- Python
- Java
- SQL
Application & API Engineering
- REST APIs
- gRPC
- FastAPI
- Microservices
- React
Identity & Security
- OAuth 2.0
- JWT
- RBAC
- IAM
Cloud & DevOps
- AWS
- Azure
- Docker
- Kubernetes
- CI/CD
- Linux
- Shell Scripting
Data Platforms & Streaming
- PostgreSQL
- MongoDB
- Redis
- Kafka
- Snowflake
- Azure Cosmos DB
- Databricks
- Data Modelling
AI & Applied AI
- Azure OpenAI
- LLMs
- RAG
- AI Agents
- Vector Search
- NL-to-SQL
- Prompt Engineering
Engineering Practices & Tools
- Git
- Agile / Scrum
- Cursor (AI-assisted development)
- Claude Code
Engineering approach
Architecture with intent.
Engineering with accountability.
Understand
Clarify user needs, business goals, constraints, risks and measurable success criteria.
Design
Shape architecture and data flows. Build AI-assisted UI prototypes to visualize workflows, gather feedback, refine the solution and define a practical MVP.
Implement
Turn the refined MVP into working software through hands-on engineering and AI-assisted development using tools like Cursor, while retaining technical ownership.
Validate
Keep validation human-led. Review AI-generated code, verify functionality through tests and edge cases, and assess security, performance and observability.
Deliver
Roll out safely, monitor production outcomes, and ensure a clean operational handover.
Contact
Let's connect.
I'm currently exploring remote senior engineering opportunities where I can lead architecture, build scalable backend platforms, and contribute hands-on to AI-driven solutions. If my experience aligns with your team's needs, I'd welcome a conversation.
Currently considering remote opportunities only.
Prefer LinkedIn? Message me there ↗