PROJECTme, mostlyBOOK No.1PAGE1DATEOctober 2026
Naman Tyagi
exhibit A

Naman Tyagi

MS Computer Science @ Northeastern University, Vancouver

systems / low-level / compilers

About

why I'm here

I'm a graduate student in Computer Science at Northeastern University's Vancouver campus. I like working close to the machine: compilers, runtimes, memory, and the gap between what a program says and what the hardware actually does.

My background is in deep learning efficiency (compression, sparsity, and optimizer benchmarking), which is what pulled me down the stack. These days I'm building tools that turn models into small, dependency-free native code.

I'm looking for internships and full-time roles in low-level / systems engineering.

Currently

  • Building Null, an ONNX → standalone C compiler, in C++20
  • Coursework at Northeastern (Fall 2026)
  • Open to systems internships for 2027

Things I like poking at

see sketch below
  • Compilers and code generation
  • Systems programming in C / C++
  • Performance engineering and profiling
  • Memory layout, caches, and SIMD
  • Emulation and computer architecture
  • ML inference runtimes
ONNXgraph IRscheduleC code!no runtime deps

fig. 1: Null, roughly

Experiments

more soon

Exp. #1: Null (2026 – ongoing)

A compiler that lowers ONNX models to standalone, dependency-free C. Written in C++20, built up one small, tested step at a time; first milestone is an MLP running end to end and matching onnxruntime.

Tools: C++20, C, ONNX

Exp. #2: ConvNet Library (Jun – Aug 2025)

A modular ConvNet library modelled on the Transformers API: config management, ResNet-style residual blocks, and a lightweight trainer for custom datasets and metrics.

Tools: Python, PyTorch, NumPy

Exp. #3: MEMO Score (Sep 2025)

A standardized benchmark for pruning, quantization and distillation across CNNs and Transformers, measuring FLOPs, latency, parameters and accuracy trade-offs. It became the basis for the MEMO paper (DASA 2025).

Tools: PyTorch, Transformers, Accelerate

Education

  • Northeastern University, Vancouver: MS, Computer Science (2026 – present)
  • Amity University: B.Tech, Computer Science and Engineering (2022 – 2026)

References

glued in
  1. Comprehensive Benchmarking of Deep Learning Optimizers: A Multi-Architecture Analysis
    V Gupta, R Lal, S Das, N Tyagi, K Kunal. 9th International Conference on Micro-electronics and Telecommunication, 2025.
  2. MEMO Benchmark: Guiding Optimization Trade-offs in Deep Neural Networks
    N Tyagi, S Das, Kunal, V Gupta, M Kakkar. International Hybrid Conference on Decision Aid Sciences and Applications, 2025.
  3. Multi-Domain Information-Theoretic Features and Kolmogorov Complexity for Lightweight Image Splicing Detection
    N Tyagi, RC Joshi, S Das, Kunal, S Vojtech, J Stepan, MK Dutta. 17th International Congress on Ultra Modern Telecommunications and Control, 2025.
  4. Reinforcement Learning for Mathematical Reasoning in Small-Scale Language Models with Structured Policy Optimization
    N Tyagi, RC Joshi, S Das, P Sikora, V Myska, MK Dutta. 17th International Congress on Ultra Modern Telecommunications and Control, 2025.
  5. Multi Stream Attention Networks with Adaptive Syntactic-Emotional Fusion for Sentiment Analysis
    N Mishra, A Gupta, S Mala, S Das, N Tyagi, S Bhardwaj, NC Debnath, .... International Journal of Computers and Their Applications, 2025.
  6. AI-Enabled Skin Disease Identification and Diagnosis Robot
    R Lal, A Sharma, Kunal, N Tyagi, S Das, V Gupta, M Singh, A Chopra, .... IN Patent 454778-001, 2025.

IF FOUND, PLEASE RETURN TO

tyagi.nam@northeastern.edu
github.com/naman466 · linkedin.com/in/naman466/
Vancouver, BC