I build the infrastructure behind robotics research: simulation environments, capture and annotation pipelines, evaluation campaigns and reproducible setups.
BSc Computer Science, BRAC University, Dhaka. Expected graduation January 2027. Available remotely from January 2027.
Computer Science undergraduate at BRAC University building the infrastructure behind robotics research: simulation environments for Isaac Sim and Unreal Engine 5/Cosys-AirSim, in-vehicle capture and annotation pipelines for a multimodal driving dataset, a scored evaluation campaign, benchmark tooling, and a pinned, checksum-verified setup script that fails unless it reproduces the reported run.
Work spans ROS 2 autonomy stacks (C++, Python), event-camera synthesis and a multi-seed training suite, backed by test suites (289 tests in hifi-environment-builder), licence gates and per-file provenance.
Sim-first terrain stack: ZED 2i depth to uncertainty-aware elevation layers, TOPP-RA profiles with path-varying limits, Isaac Sim 6 scenes for three Clearpath platforms, and a checksum-pinned setup that fails unless it reproduces the reported run.
Label-correct, sensor-clean industrial environments built with hifienv in Unreal Engine 5, with seeded layouts, QC gates and event plus RGB capture.
In-vehicle recording with INS master clock and sync validation, PPK trajectories, remote export that hashes data where it lives, and semi-automatic 3-D labelling into the LiDAR cloud.
Python toolkit and Claude Code skill that builds label-correct, sensor-clean, licence-clean, reproducible simulation environments for Unreal Engine 5 / Cosys-AirSim and Isaac Sim / USD from CC0 and CC-BY assets with provenance; parity, null-test and licence gates; 289 tests.
Sim-first research stack behind CERTUS: ZED 2i stereo depth through elevation_mapping_cupy to slope, roughness and uncertainty layers; TOPP-RA (cvxpy) velocity profiles whose limits vary along the path; Isaac Sim 6.0.1 scenes for Clearpath Husky, Jackal and Warthog; a ROS 2 Jazzy container; and a staged, idempotent setup script that pins every dependency by checksum and fails unless it reproduces the reported run.
Full classical autonomy stack for the ICRA BARN Challenge (Clearpath Jackal, 2-D LiDAR, hidden obstacle worlds): online occupancy and distance field, A*, elastic band, sequentially linearised MPC solved as a QP at 30 Hz, safety shield, recovery; ROS 2 Jazzy, C++; a 500-trial campaign protocol on the BARN 2026 public 50-world suite; a nine-chapter tutorial and a live site with seven interactive demos.
In-vehicle recording stack with supervisor and watchdog, PPK-corrected INS trajectories, a remote export pipeline that hashes hundreds of GB where they live, and a semi-automatic 3-D labelling pipeline lifting foundation-model segmentation into the LiDAR cloud. Private repository.
Event-based surveillance benchmark tooling: corrected ESIM event synthesis, class-agnostic RGB and event detector baselines, dt-aware SORT tracking and a time-to-first-detection metric; now the capture infrastructure behind COVE. Private repository.
Three-stage pipeline behind the thesis experiments: Optuna/TPE hyperparameter search, multi-seed retraining of 3 methods × 3 backbones × 8 GLUE tasks (72 combinations, 5 seeds each) through a CLI or a web portal, and an analysis stage producing mean ± std tables, single-signal ablations and Wilcoxon tests.
Listed by title only: no PDFs, and no figures, from manuscripts under review or in preparation. No author lists for unpublished work.
Five-seed, budget-matched comparison of a probe-based rank allocator against uniform-rank LoRA, LoRA and GoRA on eight commonsense tasks across three decoders plus 11 of 24 GLUE cells; the allocator trails the matched control in 26 of 35 cells.
14 h 35 min across 175 scenes from a Livox Mid-360, stereo depth, six RGB cameras and post-processed GNSS/INS, with CNG auto-rickshaws and cycle rickshaws as 3D classes and every metric reported per traffic-density level.
A rover drives only on ground it has certified: step, slope and body clearance pass with a distribution-free margin on the map's own uncertainty; a decidability rule predicts which depth sensor can certify which rover; when nothing ahead is certified, active perception picks the viewpoint that resolves the most blocking cells per second. Validated in Isaac Sim before deployment on a real rover.
The first collaborative-perception system whose shared representation is event-derived, with a counterfactually supervised, recipient-specific gate deciding what crosses the link; simulation study in Unreal Engine 5.8 / Cosys-AirSim.
Five industrial environment families with seeded layouts, exact label geometry, QC gates and rule-breaking hazard speeds, captured with event and RGB cameras.
The CV and this page say the same things; the CV has the detail.