Data Engineering & AI Pipelines.
Blending Petabyte-Scale Data with Digital Marketing & Design.
I am Praveen T, a Data & AI Pipeline Engineer with 3+ years architecting production distributed platforms. From provisioning a 19-VM HA CDC stack streaming 24 MSSQL tables in 3 days, to scaling Flipkart Ads through 21x traffic spikes, and red-teaming frontier AI models across 13 enterprise tool environments.
Scale & Production Metrics
Streaming, batch, BI, and LLM evaluation architectures
Flipkart Ads clickstream, solar IoT telemetry, and Spark batches
High-availability Kafka, Flink, and GKE deployments
Debezium, Kafka, Flink, and ClickHouse stood up in 3 days
Legacy Stored Procedures slashed from 15 min to 3 min
Dynamic auto-scaling during Big Billion Days sales events
Engineering Technologies
End-to-End Enterprise Data Flow
03. Stateful Stream Processing: Apache Flink (5 Nodes) • Beam • RocksDB
Sliding window deduplication with RocksDB state backends and Aerospike sub-millisecond cache lookups handling Flipkart Big Billion Days scale.
Featured Engineering Deployments
Frontier AI Evaluation & Multi-Tool Red-Teaming
Outlier.AI (Freelance Client)
Constructed complex adversarial scenarios across 13 mock enterprise tool environments. Authored deceptive SQL datasets and conflicting state records to probe multi-step agentic reasoning and uncover frontier model failure modes.
Near Real-Time CDC Platform for Solar Manufacturing
Tata Power Mumbai (via DataSturdy Consulting)
Architected an end-to-end cross-cloud CDC platform streaming 24 MSSQL tables across every stage of solar panel manufacturing. Provisioned a 19-VM HA cluster (Debezium, Kafka, Flink, ClickHouse) in 3 days, recognized by Tata Power leadership.
Voice to Trade: AI-Driven Execution & SEBI Compliance
Bajaj Broking (via DataSturdy Consulting)
Architected Voice2Trade for Bajaj Broking: converting call recordings into structured financial orders using Gemini 3.5 Flash and Mistral 658B, with automated SEBI regulatory consent reconciliation.
Data Engineering × Creative Design & GEO
Bridging distributed stream processing, Kafka/Flink architectures, and frontier LLM red-teaming with high-end digital marketing (SEO/AEO/GEO) and graphic design (Photoshop & Illustrator).
Data Engineering & Stream Architecture
3+ years building high-throughput pipelines. From 19-VM cross-cloud CDC topologies to RocksDB stateful deduplication on Kubernetes handling 21x traffic spikes.
Apache Spark (Dataproc / Databricks)
Production / ExpertCost/performance optimization, shuffle tuning, Dataproc cluster sizing
Apache Kafka & Google Pub/Sub
Production / ExpertClustered multi-broker HA topologies, partitioning, cross-cloud VPN transport
Apache Flink
Production / AdvancedExactly-once checkpointing, sliding windows, RocksDB state backend
Apache Beam (Java SDK)
Production / AdvancedStateful stream processing, Aerospike sub-ms cache lookup, GKE autoscaling
ClickHouse & Druid
Production / Expert9-node distributed OLAP clusters, sub-second queries on manufacturing telemetry
MSSQL Server & CDC
Production / ExpertDebezium log capture, 24 tables real-time replication, Stored Procedure tuning
Technical Summary & Q&A
What is Praveen T's data scale and distributed architecture experience?
Praveen has 3+ years architecting streaming and batch platforms at Flipkart scale, handling 21x traffic spikes with Apache Beam on GKE, stateful RocksDB deduplication, and Aerospike sub-millisecond cache lookups. In renewable energy, he designed and deployed a 19-VM high-availability CDC topology (Debezium, Kafka, Flink, ClickHouse) streaming 24 MSSQL tables in 3 days for Tata Power solar manufacturing.
How does Praveen contribute to frontier AI evaluation and red-teaming?
At Outlier.AI, Praveen designs adversarial evaluation harnesses to stress-test frontier LLM agents across 13 simulated enterprise tool environments. He authors deceptive synthetic SQL datasets to probe model reliability under contradictory state records and executes RLHF alignment across text, visual, and audio modalities for Google's Gemini.
Ready to Build Resilient Distributed Pipelines?
Available for senior Data Engineering, Streaming Architecture (Spark, Flink, Kafka), Frontier AI Red-Teaming, and hybrid technical design projects.