Last updated: 5th January 2024
Dheeraj Rajaram Reddy
Work Experience
Member of Technical Staff
July 2020 - Present
- SpotIQ: ML engine
- Cortex: Time-series forecasting
- Falcon: OLAP in-memory JIT database
- Build systems: Bazel, Gradle
- LLMs: Fine-tuning, scaling, deployment
- Production Engineering: Search + Data infra
- C++ infrastructure/core libs
- CI/CD: Jenkins, Tooling
Open Source Maintainer and Core Contributor
June 2019 - May 2020
- Implemented several core layers, optimizers, models, and the first TF2.0 native implementation of CRFs.
- Maintained the CI and package release pipelines.
Intern
January 2020 - June 2020
- Containerized execution engine.
- Query-aware data-generation.
- CI and build systems.
Intern
May 2019 - June 2019
- Rewrote a Spark data-processing pipeline to make it 3x faster and less prone to Java OOM errors.
- Added deep learning support for their automated ML platform using TensorFlow.
- Built a distributed KNN algorithm using Hybrid Spill-trees.
Team Leader and Motion-Planning Head
February 2017 - February 2019
- Lead a team of 60 engineering students, handling both the technical and managerial aspects of a robotics team.
- The team placed 1st worldwide at Intelligent Ground Vehicle Competition (IGVC) 2019.
- Won the Mahindra Rise Prize Challenge for building the best driveless car for Indian roads.
- Designed, implemented and tested the software stacks of autonomous bots and cars.
Education
Manipal Institute of Technology
2016 - 2020
B. Tech Computer Science and Engineering
- Minor in computational mathematics
Selected Projects & Contributions
- Open source: Contributions to various OSS projects, including TensorFlow, PyTorch Glow compiler, etc.
- Shadesmar: A high-performance C++ IPC library that uses shared memory for communication. Supports RPC and pub-sub.
- SummaryDB: A database written in Go for storing colossal amounts of time-series data. Uses window-based aggregations for compressing data.
- Technical blog: Authored several technical articles on programming, designs, concepts, challenging problems I’ve faced, and computers in general.
- ReiLs: A reinforcement learning framework for faster modular prototyping, deployment and benchmarking of Deep-RL algorithms. Built using TensorFlow, MPI and NCCL.
Skills
- Languages: C, C++, Java, Go, Python, Rust, JS
- Build Systems: Bazel, CMake, Gradle
- CI/CD: GitHub Actions, Jenkins
- Deep Learning: Jax, PyTorch, TensorFlow
- Parallel Programming: Arm Neon, Intel SIMD, Halide
- Distributed Computing: Docker + Kubernetes, MPI
- Miscellaneous: OpenCV, ROS, Thrift, gRPC, Protobuf
This résumé is auto-generated from a YAML file, read about it here. You can directly print this file to get a PDF.