MarTech Solutions Architect | Customer Data Platforms • Data Engineering • AI Engineering | Bengaluru, India
Solutions Architect with more than a decade of work experience across IBM, Wipro, IPG Mediabrands, and others, designing and delivering enterprise-scale customer data, marketing technology, analytics, and AI-enabled solutions. Expertise in Customer Data Platforms (Adobe RT-CDP, Braze, mParticle, Segment CDP), event-driven architectures, cloud data platforms, customer journey orchestration, identity resolution, real-time activation, and data engineering on Google Cloud and Azure.
I build open-source tools that solve practical problems around running a Customer Data Platform in production — tag/analytics implementation, audience intelligence, reverse ETL, and the data-pipeline reliability that CDPs depend on. Outside of MarTech, I also build systems-level software in Rust for fun, including an from-scratch AI-native OS kernel (the SHER OS family).
- 🔭 Current focus: Customer Data Platforms, marketing data pipelines, and the tooling that keeps them reliable
- 🤖 Exploring: Agentic AI for MarTech workflows, RAG architectures, MCP-based agent tooling
- 🎓 Background: MBA, Jansons School of Business · B.Com, Mar Ivanios College
MarTech & Customer Data Platforms Adobe RT-CDP, Braze, mParticle, Segment CDP, customer journey orchestration, identity resolution, real-time activation, Hightouch / Reverse ETL
Data Warehousing & Engineering BigQuery, Snowflake, Databricks (PySpark), Microsoft Fabric, Azure Data Factory, Airbyte / Fivetran, dbt, Apache Airflow, Apache Kafka, Apache Flink (PyFlink), InfluxDB, Telegraf, Datadog
Cloud & Data Platforms GCP, Azure, AWS · Google Cloud Storage, Azure Blob Storage / ADLS, Amazon S3, MinIO
AI — RAG & Application Development LangChain, LangGraph, LlamaIndex, Chainlit / OpenWebUI, LangSmith, DSPy, BM25 + vector similarity + reranking, Context Engineering, Ragas + DeepEval, Pinecone, Weaviate, ChromaDB, pgvector, Qdrant, Milvus
AI — Agent Automation Temporal, Langflow, n8n + Node.js, FastAPI, Firecrawl, Tavily, SerpAPI, Twilio, Claude Agent SDK, OpenAI Agent SDK, Google Agent SDK, Ollama, Scheduled & Background Agents
AI — Autonomous Code Generation Claude Code, Cursor, Replit, Codex, Hermes Agent, OpenCode, Loveable, OpenRouter
MCP & Model Platforms Model Context Protocol, Hugging Face, AWS Bedrock, Azure AI Foundry, GCP Vertex AI (Model Garden)
Infrastructure as Code & Containers Terraform, Docker, Docker Compose, Azure Kubernetes Service (AKS), NGINX Ingress
Identity & Security Entra ID / Azure AD, Azure Key Vault, Azure AI Search, incoming/outgoing guardrails
IoT, Messaging & Robotics Raspberry Pi 5, ESP32, MQTT (Eclipse / RabbitMQ), Node-RED, AWS IoT Core, AWS IoT SiteWise, CoAP, Edge Processing, Industrial IoT
Every project below links straight to its Use Cases section — the fastest way to see whether it solves a problem you actually have. Status is described plainly: if something is early-stage or has known gaps, it says so.
| Repo | What it does |
|---|---|
| PyTagManager | Crawls a site, builds a semantic DOM graph, and generates/exports analytics-tracking configs (GTM, GA4, Segment, and others) — then drives a real headless browser to verify the tracking actually fires correctly at runtime, not just that the config looks right. |
| ClusterAudienceKit | Customer segmentation engine (Rust core + Python bindings): RFM analysis, multiple clustering algorithms, churn/CLV scoring, and streaming segmentation with drift detection. |
| PyReverseETL | Moves data from warehouses/databases to CRM and marketing-automation destinations with real PII masking and a recorded lineage graph — the activation counterpart to CDP ingestion. |
| MessageBirds | Customer data platform infrastructure: Kafka-based event streaming, Postgres storage, an edge ingest gateway, and SDKs across web, iOS, and Python. |
| Repo | What it does |
|---|---|
| PyAirflowTester | Static analysis for Airflow DAGs and dbt manifests, plus dependency-intelligence (blast radius, risk scoring) and an optional web dashboard. |
| StatGuardian | Rust data-quality engine with a custom .sg validation DSL — schema validation, drift detection, anomaly detection. |
| PyBeamGuard | Static analysis for Apache Beam/Flink/Spark pipelines — flags hot keys, shuffle bottlenecks, and cost risks before deployment. |
| PyDBTGuard | Scores dbt tests for reliability, cost, and blast radius from static manifest metadata — early-stage project. |
| PyDependencyCheck | Python dependency vulnerability scanning, health scoring, and SBOM generation. |
| PyNetworkIntel | Lightweight LAN network discovery, topology mapping, and vulnerability scanning — dependency-free by design. |
| PyStreamXL | Streams large .xlsx files row-by-row in constant memory (Rust-backed), with streaming writes and formula/comment extraction. |
| PySynthData | Generates synthetic relational datasets from a schema, with differential-privacy noise injection and referential-integrity-aware row generation. |
| PyWeatherEnriched | Geocodes a location and fetches real historical weather data, then engineers climate features (rolling stats, degree-days, anomaly z-scores) for ML pipelines. |
| Repo | What it does |
|---|---|
| PyTokenCalc | Token counting and cost estimation across 10+ LLM provider tokenizers (OpenAI, Anthropic, Google, Cohere, Azure, HuggingFace, Ollama, and more). |
| PyInferenceManager | Multi-provider LLM inference orchestrator (Anthropic, OpenAI, Gemini, Ollama, vLLM) with cost tracking, retries, and circuit-breaking. |
| PyStreamMCP | Query planning and context discovery for AI agents — exposed as a Python SDK, REST API, and MCP tool server. |
| PyVectorHound | Diagnostic layer for RAG/vector-search pipelines — computes retrieval-quality metrics and root-causes failures. |
| PyStreamPDF | PDFium-backed PDF library with token-budget-aware semantic chunking, aimed at reducing what gets sent to LLMs. |
| PyAPICheck | Discovers API surface from OpenAPI/traffic, scores security risk with a named reason for every finding, and generates Cedar policy for enforcement. |
| PyStreamAI | ML deployment toolkit — canary/A-B routing, cost tracking, request scheduling, and ONNX Runtime inference. |
| PyInterviewBot | Voice-AI interview platform: Rust gateway, Python AI service, JS client — early-stage, see its own README for current limitations. |
| OpenAnchor | Token intelligence middleware for multi-provider LLM usage. |
| PyBlastRadius | Dependency-graph blast-radius analysis, criticality scoring, and failure-cascade simulation — Terraform discovery is real; Kubernetes/Airflow/OTLP sources are not yet wired up. |
| Repo | What it does |
|---|---|
| TinyBridge | macOS-native Linux VM runtime on Apple's Virtualization.framework. macOS backend is real and working; Windows/Linux backends are unimplemented scaffolding. Installed via Homebrew tap, not PyPI. |
| PrismNote | Jupyter-style data-science notebook: Rust/Axum backend, React frontend, real local SQL execution, and a Docker-sandboxed code executor. Installed via Homebrew tap. |
| MudFish | Async Rust web crawler (frontier, fetch, parser) with Python bindings via PyO3. |
A multi-repo, in-development OS project outside the MarTech day job. Most subsystems are currently a deliberate simulation layer — real Rust code with real tests, but in-memory/userspace behavior rather than real hardware I/O or a bootable kernel. Each repo's own README states plainly what's real vs. simulated.
| Repo | What it does |
|---|---|
| SHER-KERNEL | Core kernel scaffolding — object model, scheduling, memory, driver-lifecycle concepts, and a Linux Kernel Interface compatibility layer. |
| SHER-Graphics | GPU abstraction and rendering — software simulation plus a real, working Vulkan/MoltenVK backend. |
| SHER-Display | Display server, compositor, and window management. |
| SHER-INPUT | Canonical, backend-independent input event stream — real evdev backend on Linux, simulated backend for hardware-free testing. |
| SHER-Aurora | GNOME-style design system — tokens, typography, motion, and an automated WCAG contrast audit. |
| SHER-Process-Explorer | Evidence-based Linux process investigation tool — /proc telemetry, bpftrace integration, daemon + CLI + desktop UI. |
| Repo | What it does |
|---|---|
| PyRoboFrames | Rust-backed ML dataloader for robot-learning datasets — native LeRobot format support, Apple Silicon hardware video decode. |
| PyRoboReplay | Forensic-debugging and causal-analysis engine for autonomous robot mission replay. |
| PyRoboSimulator | World simulator for autonomous systems, with a FastAPI backend service — active development, see its own README for current test/CI status. |
| PyRoboVision | Multi-object tracker (Kalman filter + Hungarian algorithm) and trajectory predictor for robot perception. |
| PyTerrainMap | Fuses multi-robot terrain observations into a spatial/temporal store, with H3 spatial indexing and SLAM. |
- 📧 mullassery@gmail.com
- 📍 Bengaluru, India
- 🚀 Product Hunt
- 🔗 github.com/Mullassery
Thanks for stopping by — always happy to talk CDPs, marketing data pipelines, or the tooling that makes them reliable.