Shubham Ahuja — Portfolio

Projects

Certified

Self-directed project

Predictive Maintenance: Industrial ML for Fault Detection

Condition-Monitoring Engineer · AI / ML · July 2026

92/ 100

A ProoV case study · educational project, not employment

Built an end-to-end rolling-bearing fault detector from vibration data: visualized the signal, engineered RMS/kurtosis/crest-factor and envelope-spectrum features at the supplied defect frequencies, localized the fault to the outer race via BPFO, and trained a depth-capped decision tree on training bearings only. Evaluated on a held-out set with catch-rate and false-alarm rate, then justified a 0.50 alarm threshold using explicit miss-vs-inspection costs in a maintenance memo.

Graded against

Feature engineering (signal to honest numbers)25%
Held-out evaluation (honest measurement)25%
Cost-based threshold decision + memo25%
Compact model + embedded-budget justification15%
Plain-language explanation10%

Passed · pass mark 60/100

What stood out6

Correctly found BPFO as the dominant envelope-spectrum band and mapped it to the outer race.
Reported a compact tree with stated depth 1 and 3 nodes, aligned with the embedded-budget constraint.
Used held-out metrics with both catch-rate and false-alarm rate, then tied the threshold to explicit miss and false-alarm costs.
Wrote a maintenance memo that names the fault location and explains the tradeoff in plain language using the student's own results.
Leakage-Free Feature Engineering
Honest Held-Out Evaluation

The work I submitted20 tasks

12 tasks · 7.1k characters · Python · 8 written answers · 470 words

Initial Prediction62 characters
Example Load Plot HealthyPython · 952 characters
Load Plot Signal CodePython · 1.8k characters
Load Plot Signal205 characters
Example Rms WorkedPython · 697 characters
Time Features143 characters
Freq FeaturesPython · 1.0k characters
Fault Location239 characters
Train Classifier363 characters
Evaluation Heldout288 characters
Threshold Decision334 characters
Decision Memo1.1k characters
Initial Prediction9 words
Example Load Plot Healthy90 words
Load Plot Signal Code163 words
Load Plot Signal15 words
Example Rms Worked72 words
Time Features10 words
Freq Features97 words
Fault Location14 words

Summarised on purpose — the submitted code and writing stay private so this page cannot be reused as an answer key. The full submission sits behind the verified certificate.

Verified certificateTamper-proof · issued by ProoV

Certified

Self-directed project

Machine Learning for Automotive Safety

Perception Validation Engineer (ADAS Safety Sign-off) · AI / ML · July 2026

92/ 100

A ProoV case study · educational project, not employment

Built a perception validation dossier end to end for a fictional Aurora-7 detector: defined the IoU-based detection rule, computed precision/recall and VRU false-negative rate from a synthetic scored table, and sliced recall by lighting, visibility, and class to surface the highest-risk night and occlusion failures. Closed the loop with ISO 21448 SOTIF chains, a condition-limited ship recommendation, and a real-world comparison to the Uber Tempe crash.

Graded against

Detection metrics correctness (precision, recall, VRU false-negative rate)30%
Failure-mode analysis & per-condition rigour25%
SOTIF safety case quality (ISO 21448 reasoning)25%
Ship verdict — evidence-based & internally consistent20%

Passed · pass mark 60/100

What stood out6

Correctly locked the detection rule at IoU >= 0.5 and classified the worked pair as a miss when IoU was 0.111
Used VRU recall and VRU false-negative rate as the safety-critical metric instead of relying on aggregate accuracy
Exposed the collapse in night and occluded slices, with the worst combined slice at 0.27 recall
Wrote mitigations that are concrete ODD restrictions and fallback behaviors rather than vague monitoring language
Leakage-Free Detection Definition
Honest Safety Metric Evaluation

The work I submitted20 tasks

12 tasks · 16.3k characters · Python · 8 written answers · 687 words

Sensor Match554 characters
Task Definition377 characters
Example Metrics DemoPython · 2.2k characters
Scored Metrics RunPython · 2.3k characters
Scored Metrics351 characters
Ttcs Example SlicePython · 1.3k characters
Ttcs Triage SlicePython · 1.6k characters
Failure Triage1.7k characters
Sotif Case2.7k characters
Realcase Analysis937 characters
Teachback891 characters
Example Iou Demo1.5k characters
Sensor Match27 words
Task Definition42 words
Example Metrics Demo154 words
Scored Metrics Run143 words
Scored Metrics18 words
Ttcs Example Slice143 words
Ttcs Triage Slice112 words
Failure Triage48 words

Summarised on purpose — the submitted code and writing stay private so this page cannot be reused as an answer key. The full submission sits behind the verified certificate.

Verified certificateTamper-proof · issued by ProoV

Certified

Self-directed project

Machine Learning for Automotive

AI / ML · June 2026

92/ 100

A ProoV case study · educational project, not employment

A verified, self-directed industry project — completed and passed a real industry rubric at 92/100 on ProoV.

Graded against

Exploratory Data Analysis20%
Feature Engineering25%
Model Building & Evaluation35%
Business Communication20%

Passed · pass mark 60/100

What stood out6

Compared Linear Regression and Random Forest using multiple metrics and clearly identified the stronger model.
Interpreted the most important pricing drivers correctly: car age, mileage, and engine size.
Connected model error to business impact with a concrete portfolio-risk estimate.
Provided a practical recommendation for the CPO team rather than only reporting metrics.
Data-Driven Decision Making
Predictive Modeling

The work I submitted10 tasks

5 tasks · 6 lines · 5 written answers · 189 words

VW Sprint1 Complete56 characters
VW Sprint2 Complete61 characters
VW Sprint3 Complete59 characters
VW Business Insights1.6k characters
Jupyterlite Code2 lines
VW Sprint1 Complete1 word
VW Sprint2 Complete1 word
VW Sprint3 Complete1 word
VW Business Insights185 words
Jupyterlite Code1 word

Summarised on purpose — the submitted code and writing stay private so this page cannot be reused as an answer key. The full submission sits behind the verified certificate.

Verified certificateTamper-proof · issued by ProoV