How do we decode errors before they become computation?
Designing decoding architectures and pipelines — MWPM, Union-Find, and belief propagation — for low-latency, resource-aware FPGA deployment.
Hi, I’m Sai.
I make quantum computation more reliable and more efficient.
I started in computer science and kept chasing problems closer to the machine — until quantum hardware caught me. What I love is the narrow gap between an elegant algorithm and the silicon that has to run it on a real-time deadline, where physics, maths, and engineering are all forced to agree. I graduated my M.Tech in Quantum Technology at IISc, with distinction.
Decoding architectures+Hardware acceleration+Quantum dynamics
01 / About my work
What keeps me curious
I like the point where a clean idea meets an inconvenient machine.
I began in computer science, moving through networks, optimisation, sensing, and quantum computing. The common thread is clear: I’m happiest when I can learn very different parts of a problem and bring them together into one working system.
How fast can we simulate the dynamics themselves?
Hardware-optimised tensor-network kernels mapped onto parameterised systolic fabrics, with explicit timing, resource, and precision trade-offs.
How can experiments become reproducible tools?
Automating transmon characterisation and benchmarking quantum functions across fidelity, runtime, and noise sensitivity.
02 / Trajectory
How I got here
From algorithms on paper to algorithms in silicon.
I did not begin with a fixed plan to work on quantum hardware. I followed the problems that kept pulling me closer to the machine.
-
2019—2301 · Where it began
A computer scientist, first
At Amrita Vishwa Vidyapeetham I fell for algorithms, systems, and machine learning — the habits of thinking that still shape how I work today. I left as college gold medallist.
B.Tech Computer Science · Amritapuri · Gold Medal -
2021—2402 · Many rooms, many methods
Research in a dozen different shapes
Amrita, IIT Kanpur, Penn State, and IIIT Hyderabad took me through complex networks, sensing, scheduling, and bandit algorithms. Each lab taught me another way to ask whether an idea actually works.
Four labs · several ways of seeing -
2024—2603 · The quantum turn
IISc, and the pull toward the machine
My M.Tech drew the threads together — algorithms, physical noise, and hardware limits. An internship at IBM Quantum let me probe real backends with benchmarking and adversarial circuits. I graduated with distinction.
M.Tech Quantum Technology · IISc · Distinction -
Now04 · The work, continued
Staying for the questions that remained
I joined SINESys Lab as a Research Assistant to keep building: decoding architectures and pipelines for quantum error correction, and hardware optimisations for simulating quantum dynamics — where nanoseconds and DSP slices become part of the algorithm.
SINESys Lab · DESE · IISc
03 / Education
Where the foundations were laid
Two degrees, both finished near the top.
The numbers matter less than what they made possible — but here they are, in their proper place.
0.00/ 10 CGPA
Distinction
M.Tech, Quantum Technology
Indian Institute of Science (IISc), Bengaluru
Quantum error correction, hardware-aware algorithms, and the architectures that run them.
0.00/ 10 CGPA
Gold Medallist
B.Tech, Computer Science & Engineering
Amrita Vishwa Vidyapeetham, Amritapuri
Algorithms, systems, networks, and machine learning — the grounding for everything since.
Earlier
Intermediate · FIITJEE Jr College, Vizag · 89%
Schooling · SR Digi School, Vizag · 10.0 CGPA
04 / Skills
What I actually do
A few things I do deeply — and a long tail of tools behind them.
Quantum Error Correction
Surface codes, syndrome extraction, and logical-error analysis with Stim.
Decoding Architectures
MWPM, Union-Find & belief propagation — as algorithms and as pipelines.
FPGA & Hardware Acceleration
Verilog RTL, Vitis HLS, ASIC/FPGA flows, timing- and resource-aware design.
Quantum Algorithms & Simulation
Tensor-network dynamics, Qiskit, and hardware-optimised simulation kernels.
Algorithms & Optimisation
Randomised & approximation algorithms, scheduling, bandits, complexity.
Machine Learning
Calibration, sensing, and link prediction with TensorFlow & scikit-learn.
Languages
Python · C / C++ · Java · Scala · Haskell · SQL
Hardware & EDA
Verilog · Vitis HLS · Vivado · LLVM / MLIR · 45 nm ASIC
Quantum
Qiskit · Stim · PyVISA · cavity-QED fitting
ML & Web
TensorFlow · Keras · scikit-learn · Django · React · PyQt6 · REST
05 / Two theses, one instinct
The pattern I recognise now
I like bringing many kinds of thinking to one problem.
My two theses look unrelated at first: one is quantum decoder hardware, the other a social movie game. What connects them is the kind of work I gravitate toward — crossing layers, comparing approaches, and making the pieces meet in a complete system.
2026
M.Tech
Individual thesis · Quantum Technology · Indian Institute of Science · Distinction
Hardware Acceleration for Quantum Decoding
I studied surface-code decoding from two connected viewpoints: whether an algorithm suppresses logical errors, and whether its hardware can meet a real-time latency budget.
What it brought together
MWPM, Union-Find, and belief propagation · Stim-based logical-error studies · profiling-driven kernel selection · verified Verilog RTL · 45 nm ASIC synthesis · ZCU104 FPGA implementation · Vitis HLS comparison
The central lesson: accuracy and speed must be measured separately, then combined honestly. Fewer cycles do not mean faster hardware if the critical path destroys the clock.
Read the M.Tech thesis ↗
2023
B.Tech
Team thesis · Computer Science & Engineering · Amrita Vishwa Vidyapeetham
Guess It! An Online Multiplayer Trivia
We began with a simple human goal: help friends have a good time together. Building the game meant creating the entire path from raw movie data to a playable web experience.
What it brought together
Web scraping and public datasets · NLP keyword-extractor benchmarking · image processing and object detection · similarity mapping · Django application architecture · cloud deployment
What I learned: integration is its own kind of problem solving. A clever component matters only when it fits the data, the interface, and the person using it.
Read the B.Tech thesis ↗06 / Research
What I’ve been building
The research, in its own proper context.
Results stay with the work that produced them. This is where the numbers, trade-offs, and technical detail belong.
SINESys Lab · DESE · IISc
Decoding architectures & pipelines for real-time QEC
A combined algorithm-and-hardware study of MWPM, Union-Find, and belief-propagation decoders across logical error rate, wall-clock latency, code-distance scaling, and hardware cost — built as reusable pipelines from syndrome to correction.
- Golden-model verified RTL
- ASIC + FPGA + HLS flows
- Accuracy and wall-clock speed
Course project · IISc
Systolic fabric for tensor-network quantum dynamics
Parameterised 4×4 to 32×32 GEMM fabric on Xilinx ZCU104, joining Verilog compute cores to a Vitis HLS tensor-network layer for hardware-accelerated dynamics.
569.5MHz Fmax
14.8%DSP use
82—6500×modelled speed-up
IBM Quantum
Benchmarking & adversarial circuits
An implementation-benchmarking suite for Qiskit Functions, and an evolutionary search that breeds adversarial circuits to maximise error across hardware backends.
Experience record
Built across disciplines, always measured against reality.
Quantum Computing Intern · IBM
Implementation-benchmarking suite for Qiskit Functions; evolutionary search for adversarial circuits that maximise error across hardware backends.
Research Intern · IIIT Hyderabad
Multi-objective scheduling theory and delay-aware multi-arm bandit optimisation for a large recommendation system.
Research Intern · Penn State
Deep-learning calibration for 2D-semiconductor chemical sensors; neuromorphic CSP algorithms for 2D devices.
SURGE Research Fellow · IIT Kanpur
Quicksort implementation profiling and a stable-matching algorithm for departmental TA allocation.
Project Assistant · Amrita
Link prediction with SBMs and MLE in complex networks; influential-node identification via mPageRank.
07 / Projects
Things I’ve built
Projects where an idea became something that runs.
Beyond the research — tools, hackathon builds, and applications taken end-to-end.
QTL · IISc · Research tool●
TRACE-Q
Transmon Reliability, Automated Characterisation & Estimation for Qubits. A lab-instrument GUI that fits VNA/IQ data to extract internal quality factor, anharmonicity, and dispersive shifts automatically.
PyVISAPyQt6Cavity QEDNumPy/SciPy
Smart India Hackathon 2025↗
QuMail
A Windows desktop client that layers Quantum Key Distribution over ordinary email. Application-layer encryption for messages and attachments with REST-based key management — post-quantum secure mail.
PythonQKDTkinterREST API
Web app · B.Tech thesis↗
Guess It!
An online multiplayer trivia where players guess movies from generated image clues. Full pipeline: web-scraped datasets, NLP keyword extraction, similarity mapping, a Django backend, and a React front end.
DjangoReactJSRESTNLP
Networks · Independent research●
Link Prediction & Motifs
Predicting links in directed and multiplex social networks using stochastic block models, MAP estimation over motifs, and m-PageRank influence clustering — the basis of three publications.
PythonSBM / MLENetworkXscikit-learn
Also built along the way
Attendance Management System · Tkinter + SQLite
Fantasy Cricket Game · PyQt5
Event Management System · Java + PostgreSQL
Personality Assessment · EDA + clustering
NFC Card Authentication · LPC2148 + KEIL
Anime Wiki · HTML/CSS/JS
08 / Live on Google Cloud
Shipped & online right now
Two apps I run live on GCP.
Self-hosted side projects — deployed, public, and serving traffic as you read this.
Live · Cloud Run
Scriptorium
A medieval-styled research journal — projects, journal entries, tasks, meetings, and materials, bound together like an illuminated manuscript for modern research.
Cloud Run · Web appOpen ↗
Live · Firebase Hosting
Solo Learner — The System
A “Solo Leveling”-inspired learning RPG. Turn studying into quests, gain XP, level up, and watch your stats climb — a void-black, gamified system for self-driven learning.
Firebase · Static SPAOpen ↗
This very site — nasapakri.com — hand-built and self-hosted on Cloud Storage with an HTTPS load balancer, managed SSL, and a global CDN.
nasapakri.com ↗09 / Milestones
A few things I’m proud of
Milestones, kept separate from the research itself.
These mark parts of the journey. They are not substitutes for the work above.
GATE-CS 2024
The rank that brought me to IISc and into quantum technology.
Global · LeetCode contest rating
A long-standing love for competitive algorithms and problem solving.
Regionalist
Among the teams advancing through the ACM-ICPC regionals.
Research Fellow · IIT Kanpur
My first immersive summer research experience.
10 / Publications
Writing & evidence
Ideas that leave the notebook.
Peer-reviewed work in complex networks, information extraction, and directed motifs.
01↗
02↗
BDCC · 2023 · 7(1), 31
An Improved Link Prediction Approach for Directed Complex Networks Using Stochastic Block Modeling
PiCET · 2024
Surveying Keyword Extractors: Classification, Applications, and Empirical Analysis
03Accepted
ISTA · 2023 · Accepted
Probabilistic Distributions on Directed Motifs in Complex Social Networks: A Deep Analysis
11 / Teaching & service
Research is a community practice
I teach, convene, and make room for other people’s questions.
QT 207 · Introduction to Quantum Computation
Teaching Assistant at IISc, supporting tutorials and mentoring students through the mathematical foundations of quantum computation.
Aug 2025—PresentIISc × Qiskit Fall Fest 2025
Led a five-member team across speaker invitations, session delivery, logistics, and hackathon judging for the IBM Quantum / IQTI / QuRP-sponsored event.
Lead organiser · 2025Department representative & seminar coordinator
Representing the M.Tech cohort, running the IAP seminar series, and coordinating student-wellness initiatives through SWAN.
IISc · 2024—PresentIf our questions overlap
I’d love to
hear from you.
I’m always happy to talk about quantum error correction, decoding architectures, hardware-aware algorithms, FPGA acceleration, teaching, or a possible research collaboration.