Abdulmalik AjisegiriAbdulmalik Ajisegiri is a systems engineering and model risk professional specializing in model validation, AI/ML governance, advanced analytics, and MBSE.https://abdulmalikajisegiri.com/Validating Models Like a Skeptic: The Outcomes-Analysis Playbookhttps://abdulmalikajisegiri.com/research/validating-models-like-a-skeptic/https://abdulmalikajisegiri.com/research/validating-models-like-a-skeptic/The SR 11-7 validation playbook: conceptual soundness, outcomes analysis with challenger models, ongoing monitoring — the attempt to kill your model.Sun, 20 Sep 2026 00:00:00 GMTBuilding an LLM Evaluation Harness: BLEU, ROUGE, SBERT, and Risk Tagginghttps://abdulmalikajisegiri.com/research/llm-evaluation-harness-bleu-rouge-sbert/https://abdulmalikajisegiri.com/research/llm-evaluation-harness-bleu-rouge-sbert/A repeatable LLM evaluation harness: what BLEU, ROUGE, and SBERT actually measure, where each misleads, and the architecture turning scores to governance.Wed, 16 Sep 2026 00:00:00 GMTStacking Ensembles That Actually Generalize: Lessons from Clinical Risk Modelinghttps://abdulmalikajisegiri.com/research/stacking-ensembles-clinical-risk-modeling/https://abdulmalikajisegiri.com/research/stacking-ensembles-clinical-risk-modeling/Out-of-fold stacking discipline and validation skepticism that keep ensemble gains honest — R² 0.78 base model vs 0.81 stacked, on a clinical risk project.Fri, 11 Sep 2026 00:00:00 GMTWhen R² Lies: Evaluating Regression for High-Stakes Decisionshttps://abdulmalikajisegiri.com/research/when-r2-lies-regression-evaluation/https://abdulmalikajisegiri.com/research/when-r2-lies-regression-evaluation/Why R² can’t tell you a clinical model is safe: calibration, cost-aware thresholds, subgroup slices, and honest uncertainty beyond the leaderboard number.Mon, 07 Sep 2026 00:00:00 GMTSurvivorship Bias and Lookahead: Two Backtest Killers, with Codehttps://abdulmalikajisegiri.com/research/survivorship-bias-lookahead-backtest-killers/https://abdulmalikajisegiri.com/research/survivorship-bias-lookahead-backtest-killers/Two classic backtest bugs in runnable numpy: survivorship bias tests only assets that exist today; lookahead bias trades on unknowable information.Wed, 02 Sep 2026 00:00:00 GMTMonte Carlo for Decisions Under Uncertaintyhttps://abdulmalikajisegiri.com/research/monte-carlo-decisions-under-uncertainty/https://abdulmalikajisegiri.com/research/monte-carlo-decisions-under-uncertainty/When simulation beats closed-form math: the 1/√N error tax, antithetic variates and common random numbers for variance reduction, and honest reporting.Sat, 29 Aug 2026 00:00:00 GMTDiscrete-Event Simulation Experiments: Replications and Honest Statisticshttps://abdulmalikajisegiri.com/research/discrete-event-simulation-experiment-design/https://abdulmalikajisegiri.com/research/discrete-event-simulation-experiment-design/An Arena poker-chip simulation as an experiment-design case study: independent replications, paired comparisons with common random numbers, intervals.Mon, 24 Aug 2026 00:00:00 GMTAngle of Arrival on a Budget: Localizing Sound with Four Microphones and a Cortex-M4https://abdulmalikajisegiri.com/research/angle-of-arrival-embedded-audio-localization/https://abdulmalikajisegiri.com/research/angle-of-arrival-embedded-audio-localization/Localizing sound with four microphones and a Cortex-M4: time-difference-of-arrival, GCC-PHAT cross-correlation, DMA double-buffering, and honest limits.Thu, 20 Aug 2026 00:00:00 GMTDMA Double-Buffering for Continuous Sensor Capturehttps://abdulmalikajisegiri.com/research/dma-double-buffering-sensor-capture/https://abdulmalikajisegiri.com/research/dma-double-buffering-sensor-capture/Why per-sample interrupts collapse at high rates, and how DMA ping-pong keeps sensor streams gapless: buffer sizing, overrun math, cache coherency.Sat, 15 Aug 2026 00:00:00 GMTLow-Power TM4C123 Design: Hibernation, EEPROM, and a Validating UART CLIhttps://abdulmalikajisegiri.com/research/low-power-embedded-design-tm4c123/https://abdulmalikajisegiri.com/research/low-power-embedded-design-tm4c123/An unattended weekend pet feeder on the TM4C123: RTC-wake hibernation, EEPROM schedules with wear budgets, a validating UART CLI, comparator water sensing.Tue, 11 Aug 2026 00:00:00 GMTHow System Calls Actually Work: A Tour of a Tiny C Kernelhttps://abdulmalikajisegiri.com/research/how-system-calls-work-tiny-c-kernel/https://abdulmalikajisegiri.com/research/how-system-calls-work-tiny-c-kernel/Tracing a userland call like cat through the trap, dispatch table, and syscall implementations of a tiny C kernel — the user/kernel boundary demystified.Thu, 06 Aug 2026 00:00:00 GMTZod at the Boundary: Validating Every Byte That Enters Your APIhttps://abdulmalikajisegiri.com/research/zod-at-the-boundary-api-validation/https://abdulmalikajisegiri.com/research/zod-at-the-boundary-api-validation/TypeScript types evaporate at runtime: a Zod-first Express design for the trust boundary — validate-once middleware, clean 400s, coercion, fuzz testing.Sat, 01 Aug 2026 00:00:00 GMTCodifying Engineering Judgment: What an Automated Code-Review Pipeline Should Actually Checkhttps://abdulmalikajisegiri.com/research/automated-code-review-pipeline/https://abdulmalikajisegiri.com/research/automated-code-review-pipeline/The three layers of automated code review, how to codify engineering judgment as checkable rules, and the line automation must not cross.Mon, 27 Jul 2026 00:00:00 GMTVerification and Validation Across the Systems Lifecyclehttps://abdulmalikajisegiri.com/research/verification-validation-systems-lifecycle/https://abdulmalikajisegiri.com/research/verification-validation-systems-lifecycle/V&V isn’t a phase, it’s a thread: building right vs. building the right thing, the four verification methods, and calibrating rigor to consequence.Wed, 22 Jul 2026 00:00:00 GMTFrom Stakeholder Needs to Verification: Requirements Engineering for Complex Systemshttps://abdulmalikajisegiri.com/research/requirements-engineering-complex-systems/https://abdulmalikajisegiri.com/research/requirements-engineering-complex-systems/From stakeholder needs to verification via a toy parcel-sorting facility: elicitation, verifiable requirements, decomposition, change, end-to-end tracing.Sat, 18 Jul 2026 00:00:00 GMTUsing Optimization and Metaheuristics for Engineering Decision Problemshttps://abdulmalikajisegiri.com/research/optimization-metaheuristics-decisions/https://abdulmalikajisegiri.com/research/optimization-metaheuristics-decisions/When gradients stall on messy trade spaces: simulated annealing, genetic algorithms, particle swarm on synthetic toys, compared under an honest protocol.Mon, 13 Jul 2026 00:00:00 GMTModel-Based Systems Engineering and Requirements Traceabilityhttps://abdulmalikajisegiri.com/research/mbse-requirements-traceability/https://abdulmalikajisegiri.com/research/mbse-requirements-traceability/Traceability is what makes MBSE work: the needs-to-verification chain, typed links, trace matrices, change-impact analysis, kept alive in real projects.Thu, 09 Jul 2026 00:00:00 GMTBacktesting Quantitative Strategies Without Fooling Yourselfhttps://abdulmalikajisegiri.com/research/backtesting-without-fooling-yourself/https://abdulmalikajisegiri.com/research/backtesting-without-fooling-yourself/The ways backtests lie — lookahead bias, survivorship bias, costs, overfitting — shown in runnable Python, plus a walk-forward workflow that catches each.Sat, 04 Jul 2026 00:00:00 GMTDetecting Model Decay: Ongoing Monitoring in Productionhttps://abdulmalikajisegiri.com/research/detecting-model-decay-monitoring/https://abdulmalikajisegiri.com/research/detecting-model-decay-monitoring/Monitoring that catches decay before stakeholders do: PSI and drift checks, signal-vs-noise tracking, retraining triggers, alerts without pager fatigue.Tue, 30 Jun 2026 00:00:00 GMTStress Testing and Scenario Analysis for Quantitative Modelshttps://abdulmalikajisegiri.com/research/stress-testing-scenario-analysis/https://abdulmalikajisegiri.com/research/stress-testing-scenario-analysis/Designing historical, hypothetical, and reverse stress scenarios that break your model on purpose before reality does: severity ladders and governance.Thu, 25 Jun 2026 00:00:00 GMTAgentic AI Model Risk Management Platformhttps://abdulmalikajisegiri.com/projects/agentic-mrm-platform/https://abdulmalikajisegiri.com/projects/agentic-mrm-platform/A personal project: an agentic platform that automates the model-risk identification lifecycle — SR 11-7 classification, validation artifact generation, and pipeline tracking with SLA enforcement and audit trails.Sun, 01 Feb 2026 00:00:00 GMTClinical Risk Predictive Model — Cardiac Surgery Outcomeshttps://abdulmalikajisegiri.com/projects/cardiac-surgery-risk-model/https://abdulmalikajisegiri.com/projects/cardiac-surgery-risk-model/A stacked-ensemble ML pipeline predicting operative mortality in cardiac surgery from preoperative variables, built against STS-aligned outcomes.Sun, 01 Jun 2025 00:00:00 GMTQuantEdge Labshttps://abdulmalikajisegiri.com/projects/quantedge-labs/https://abdulmalikajisegiri.com/projects/quantedge-labs/An end-to-end trading research platform — ML-calibrated opportunity scoring, options analytics, and simulation-based verification, shipped as a full-stack web terminal.Wed, 15 Jan 2025 00:00:00 GMT