Mahmudul Faisal
Al Ameen

Shifa

Live Symbolic heap-based information-flow analyzer

Shifa proves non-interference for C programs: that a function's public outputs cannot depend on its secret inputs. It combines separation-logic reasoning about the heap with information-flow analysis over a Z3 backend, checking each function against a lightweight stub specification of its security signature (Hi → Lo).

• In-browser playground — write C, get a verdict
• Information-flow mode with Hi/Lo security labels
• Stub specification language for compositional analysis
• Separation logic + symbolic execution + Z3
• BinRevZ — reach binaries via LLVM reverse engineering

SeenSlide

In development Real-time study & presentation platform

Navigate presentations. Study together. In real-time. SeenSlide captures slides live during lectures and conferences and syncs them to every device in the room — no app, just a link or QR code. Around that live feed it adds a collaborative study layer: shared annotations, slide-linked chat, study groups, and an AI assistant that can explain a selected region, summarize a slide, or translate it on the spot.

• Live slide capture & WebSocket sync
• Collaborative annotation & chat
• AI personas — Strict Professor, Socratic Guide, ELI5…
• Select-region explain · summarize · translate
• Study groups with join codes · QR sharing
• Open-source desktop capture · cloud SaaS

Ostudi

In development ওস্তাদি · a classroom in your phone

Take live classes. Learn from wherever you are. Ostudi turns an ordinary phone into a complete online classroom — show slides, write on a live board, talk in real time, take attendance, and lean on an AI helper. No laptop, no expensive software, and it holds up on weak connections. Built for teachers and students in exactly the conditions most edtech ignores.

• Live slides, board & real-time voice
• Attendance and an in-class AI assistant
• Designed for low bandwidth & modest phones
• Android 8.0+ · free · multilingual (বাংলা · EN · FR)

Banglar Rokto

Live বাংলার রক্ত · open memorial archive

Blood of Bengal — from '52 to '24. An open, Bengali-language archive of those who gave their lives for the freedom and rights of the people of Bangladesh — spanning from the 1952 Language Movement to the 2024 student uprising. A work of remembrance, built so these names and stories cannot be quietly erased.

• Decentralised storage on IPFS — censorship-resistant by design
• Interactive timeline (1952 → 2024) & map of events
• Narrative chapters & victim profiles in Bangla
• Open archive · React · D3 · Leaflet

SeekSync

In development digital backbone for migrant labour

Not merely sending workers abroad — building a trusted human-excellence export hub. SeekSync is an operational prototype (BGCDP 3.1) for Bangladesh's overseas-labour ecosystem: an 8-layer pipeline that connects workers, training providers, recruiters, foreign employers, returnees, and the ministry around a single, dignified digital identity for every worker.

• Worker Digital Identity — NID, skills, contracts, wages, grievance & return in one thread
• AI demand intelligence — corridor-by-corridor labour forecasting
• Role dashboards: worker, provider, recruiter, employer, returnee, ministry
• Built around migration corridors: Japan, Saudi Arabia, Italy, Germany, Korea

SafeSelf

In development Privacy research & platform

My most ambitious project: adversary-aware privacy under an explicit LLM-amplified adversary threat model. SafeSelf measures how well intelligent analyzers — cloud LLMs, local LLMs, even humans — can reconstruct a person from their OSINT footprint, and how effectively decoy interventions degrade that reconstruction without making the footprint implausible. Where most tools just hide or delete data, SafeSelf attacks the adversary's confidence itself.

• seeself — personal data-audit pipeline
• safesocial — peer-to-peer social network
• counterself — end-user defensive walker
• Decoy-validation research: personas, synthetic corpora, cross-analyzer benchmark

Spheres

In development Your circle. Your peace. Your privacy.

A social network built for calm, meaningful connection — no rage-bait algorithms, no ads mining your attention, no strangers in your feed. Just the people you actually trust. Spheres is peer-to-peer: your data stays on your device, and metadata stays private by construction.

• Peer-to-peer — your device is the storage, not a company server
• End-to-end encrypted, no plaintext metadata
• No phone or email required to join
• A feed of only the people you chose — no algorithm