Santa Clara, California, United States
[No Fancy awards that I can post here]
I’m your average engineer who fails often but learns quickly.
Past experiences include building MLOps infra for batch and continuous data ingestion, pub/sub, model training, inferencing and monitoring.
Some fun facts about me - I have failed interviews at least twice in every FAANG company (some even more).
Experiences include working with ancient tech, great tech, and terrible tech software stacks.
Previously, I was:
1. Head of adamance: Demonstrated "issue is with your setup, not the app"
2. PM of time tables: Still missed morning meetings fairly often
3. Tech Lead who made major contributions: By replacing
o Led a team of 4 engineers to build an Inferencing platform to enable quick feature discovery and reduce turn around time for customer demos. o Architected a two-tier memory, utilizing Redis Streams for short-term scratchpad memory and Postgres for persisting user preferences. o Engineered the orchestrator, domain supervisors, and sub-agents (swarm) workflow using LangGraph that executed dynamic, and predefined directed acyclic graphs (DAGs) on demand. o Built an image similarity service (using Clickhouse as the vector store) that the swarm graph could interface with to find similar coupon defects or patterns, enhancing engineers’ productivity.
o Built container monitoring, log aggregation, load balancing, and autoscaling platform components ensuring high availability of the Applied AIx analytics product line for inferencing. o Developed the continuous data ingestion pipeline (using gRPC) for sampling high frequency sensor data for wafer non-uniformity analysis algos. o Engineered and deployed the batch data ingestion workflow (a multi-threaded pub/sub application for parquet transfers) for storing context, ground truths, and post-run data on the AIx on-prem cloud. o Set up the feature engineering pipeline for spectral data pre-processing and created the logic for the default transformations of PCA, FFT and ICA. o Championed ML practitioner productivity by creating the sdk for containerized algorithm deployment and defining the deployment strategy (scheduled/triggered/long-lived workflows). o Automated production software release by setting up CI/CD pipelines using Python dependency managers, Docker, Kubernetes, Helm, and Argo, ensuring streamlined and automated deployments. o Developed the ML monitoring service, and experiment manager platform for model/data evaluation, to support experiment tracking, reproducibility, model versioning, and artifact management using MLFlow, S3 and PostGres. o Led a team of 2 junior engineers to build a feature store (tenant provisioning, automatic schema healing), and training and post-processing workflow templates which were crucial in achieving automated run-to-run control during the etch fabrication step. o Led a team of 6 engineers to develop security infrastructure for authenticated TLS communication, protecting data at rest using AES encryption, and generating feedback signatures with HMAC keys, using PKCS11 compliant Hardware Security Modules (HSMs) for storing assets.
o Developed and validated exposure algorithms, calibration methods, and diagnostic routines for a Digital Lithography tool used in OLED display patterning. (tech stack was all C99) o Collaborated with field engineers to troubleshoot and resolve software issues at customer sites, ensuring system reliability. o Enhanced dose sensor and camera focus calibration software. o Optimized exposure path to reduce takt time, improving overall efficiency.
•Researched about state of the art vector quantisation techniques used to speedup similarity search. •Developed ANN’s based on this research with triplet loss functions. •Used hamming and Euclidean distance, and Discounted Cumulative Gain (DCG) to compute ranking accuracy.
•Analysed the financial behaviour of urban Indians and identified trends. •Compiled data and designed an in-depth research report displaying both data and our analysis. •Identified potential business opportunities for the BFSI sector on the basis of our analytics. •Planned the marketing and sales strategy. •Explained the benefits of the product to more than 20+ CxOs and midlevel executives of this sector. •Spearheaded the new vertical to achieve its financial targets. This experience helped me develop my interpersonal and organisational skills.