$ whoami

shreyansh soni

msc computer science student at trinity college dublin, with a background in systems, backend engineering, and a growing obsession with real-time graphics. i spent three years at deloitte shipping production-grade backend systems and gen-ai tooling — now i'm deep in the low-level end: writing shaders, building game engines, and understanding what actually happens between the cpu and the screen. i like things that are fast, correct, and a little bit weird.

01. experience

may 2025 – aug 2025 consultant

deloitte bengaluru, india

  • led a team of 3 engineers to deliver an end-to-end report creation tool
  • reduced a 4-week manual process down to a 15-minute automated job
aug 2022 – may 2025 analyst

deloitte bengaluru, india

  • delivered a gen-ai chatbot for the branch portal at wells fargo
  • built a rainy-day scenario flow keeping the bot responsive in edge cases
  • created an automated testing framework scaling 900+ queries in under 3 minutes
  • built backend & frontend microservices with java spring boot, js & jsp
jan 2022 – jul 2022 intern

siemens eda noida, india

  • created 5 custom linux distributions to spec
  • implemented dts manipulation on exynos 9 boards
  • wrote kunit test cases for an i2c serializer driver using fff

02. education

msc computer science
trinity college dublin
2025 – 2026
real-time systems graphics distributed systems xr / vr
be electronics and communication engineering
thapar institute, patiala
2018 – 2022
algorithms os networks ml

03. skills

graphics & rendering

opengl glsl unreal engine 5 hlsl rasterization ray tracing pbr npr rendering normal mapping mipmapping shadow mapping deferred rendering gaussian splatting photogrammetry

systems & low-level

c++ c linux kernel embedded (esp32) iot gossip protocols distributed systems memory management dsp

backend & cloud

java spring boot microservices rest apis mongodb sql aws gcp nodejs python ci/cd

tools & workflow

git emacs arch linux hyprland pytorch (rocm) meshroom unity cmake docker

04. currently

learning:
  - non-photorealistic rendering techniques
  - quaternion-based inverse kinematics (fabrik)
  - gaussian splatting + 3d reconstruction
  - advanced emacs configuration

building:
  - real-time opengl renderer
  - iot gossip protocol system (esp32)
  - dungeon runner in unreal engine 5

listening_to: "probably something on low-level graphics"
os: "arch linux + hyprland (btw)"

05. publications

dynamically adaptive switching based median mean filter for removal of salt and pepper noise

international conference on ai and big data in digital era · 2021–2022

design of a multipurpose android-controlled robotic arm for a smart city towards smart world

2020