Skip to content
SwankyForge
SwankyForge
Work

Shipped systems.

Problem, metric, result. The long write-up is one click in.

Product AnalyticsExperimentation6 months

Masters

Four analytics tools disagreed about the same numbers. Now there is one set everyone trusts.

Offer CTR — before
4.5%
After
6.4%
Read the write-up
SaaSData Pipelines4 months

Plug-and-Play Commerce Analytics

Signing up a new customer took three weeks of engineering time. It now takes two days and no engineer.

Tenant onboarding — before
3 weeks
After
2 days
Read the write-up
NLPVector Search3 months

Semantic Deduplication at Ingestion

The same story kept arriving five times over. Now it arrives once.

Duplicate pair F1 — before
0.65
After
0.94
Read the write-up
ASRNLP3 months

Jet Tracker

Analysts spent their shifts typing up radio traffic. Now they interpret it instead.

Segments processed / day — before
640
After
24,000
Read the write-up
Data PipelinesMetrics3 months

USDDub

Dashboards were six hours behind reality. Now they are five minutes behind.

Data freshness — before
6 hours
After
5 minutes
Read the write-up
LLMRAG4 months

Document Intelligence (OCR, Extraction, RAG)

Every invoice was read and typed in by hand. Four in five now go through untouched.

Straight-through processing — before
0%
After
81%
Read the write-up
NLPAutomation2 months

Support Ticket Triage + Auto-Routing

A third of support tickets went to the wrong team. Now it is closer to one in thirteen.

Routing accuracy — before
65.3%
After
92.4%
Read the write-up
ML SystemsMLOps5 months

Fraud Detection Pipeline + Monitoring

Analysts spent their day on false alarms. The review queue is now a third of the size.

False positive rate — before
4.1%
After
1.6%
Read the write-up
ForecastingOptimization4 months

Demand Forecasting + Inventory Optimization

Planning ran on a spreadsheet and one growth rate. Stock now follows actual demand.

Weighted MAPE — before
12.1%
After
2.7%
Read the write-up
LLMRAG4 months

Internal LLM Copilot (Policies + Tools)

Staff asked colleagues instead of reading the policy. Now they ask and get an answer they can check.

Answer accuracy — before
61%
After
89.2%
Read the write-up

Need a system like one of these?

Start from a calculator template or send the outcome.