AI Food Traceability System

We encountered a situation where a recall of dairy products affected 15% of the market—every hour of downtime cost the manufacturer millions of rubles. The lack of pinpoint traceability turned a local defect into a large-scale crisis. With AI and graph analysis, the time to locate a problematic lot is reduced from weeks to 15 minutes. Savings on a single such incident are up to 95% of recall costs. Below is how we build such systems and what problems we solve.

Problems We Solve

Mass recalls due to inaccurate localization. Traditional one-up/one-down accounting does not allow quick building of a complete graph of affected lots. An AI agent automatically collects data from all supply chain participants and calculates the risk subset in seconds. Graph analysis based on NetworkX processes a chain of 10 million nodes 100 times faster than SQL queries.

Manual labeling and input errors. When integrating with Chestny Znak, DataMatrix codes need to be applied, entry into circulation confirmed, and reconciled with physical presence. We automate request generation, code reception, and reconciliation, eliminating human error. Discrepancies drop from 5–10% to 0.5%.

Unaccounted cold chain breaches. Transport and storage with temperature violations reduce shelf life. Our ML model calculates Mean Kinetic Temperature (MKT) from IoT logger data and recalculates the remaining shelf life of each lot in real time. (MKT calculation per ICH Q1A standard)

Traceability Architecture

The graph links all participants—from raw material supplier to store shelf. The code below implements basic classes for recording a lot and building the graph.

from datetime import datetime
import json
import hashlib

class TraceabilityRecord:
    """Запись прослеживаемости для партии продукта"""

    def __init__(self, lot_id, product_code, quantity_kg, timestamp=None):
        self.lot_id = lot_id
        self.product_code = product_code
        self.quantity_kg = quantity_kg
        self.timestamp = timestamp or datetime.now().isoformat()
        self.inputs = []          # из каких партий сделан (сырьё)
        self.processing_params = {}  # производственные параметры
        self.outputs = []         # в какие партии ушёл (полуфабрикат, готовая)
        self.shipments = []       # кому отгружен

    def add_input_lot(self, input_lot_id, quantity_used, quality_params=None):
        """Добавить входящую партию сырья"""
        self.inputs.append({
            'lot_id': input_lot_id,
            'quantity_kg': quantity_used,
            'quality': quality_params or {},
            'timestamp': datetime.now().isoformat()
        })

    def record_processing(self, params):
        """Записать производственные параметры"""
        self.processing_params = {
            **params,
            'recorded_at': datetime.now().isoformat()
        }

    def add_shipment(self, destination_id, quantity, transport_conditions=None):
        """Записать отгрузку"""
        self.shipments.append({
            'destination': destination_id,
            'quantity_kg': quantity,
            'transport': transport_conditions or {},
            'timestamp': datetime.now().isoformat()
        })

    def generate_lot_hash(self):
        """Хэш записи для верификации целостности"""
        data = json.dumps({
            'lot_id': self.lot_id,
            'inputs': self.inputs,
            'processing': self.processing_params
        }, sort_keys=True)
        return hashlib.sha256(data.encode()).hexdigest()


class TraceabilityGraph:
    """Граф прослеживаемости для анализа и отзывов"""

    def __init__(self):
        import networkx as nx
        self.graph = nx.DiGraph()

    def add_lot(self, lot: TraceabilityRecord):
        self.graph.add_node(lot.lot_id, data=lot.__dict__)
        for inp in lot.inputs:
            self.graph.add_edge(inp['lot_id'], lot.lot_id, quantity=inp['quantity_kg'])
        for shipment in lot.shipments:
            self.graph.add_edge(lot.lot_id, shipment['destination'], quantity=shipment['quantity_kg'])

    def recall_simulation(self, problem_lot_id):
        """Выявить все партии, затронутые отзывом"""
        import networkx as nx
        # Все потомки (вниз по цепи): куда ушёл проблемный продукт
        downstream = nx.descendants(self.graph, problem_lot_id)
        # Все предки (вверх по цепи): из какого сырья сделан
        upstream = nx.ancestors(self.graph, problem_lot_id)

        return {
            'problem_lot': problem_lot_id,
            'downstream_lots': list(downstream),
            'upstream_lots': list(upstream),
            'total_kg_at_risk': sum(
                self.graph.nodes[lot]['data']['quantity_kg']
                for lot in downstream
                if lot in self.graph.nodes
            )
        }

How AI Connects the Supply Chain?

Each participant—from farmer to distributor—submits their data to the graph via API or file uploads. An ML model cleans and normalizes records, automatically detecting duplicates and inconsistencies. The graph is then analyzed for cycles (closed loops) and isolated nodes, which often indicate data gaps. This provides a complete picture of product movement. Graph build time for 10 million nodes is under 3 seconds (p99).

Why is Chestny Znak Integration Critical?

According to regulatory requirements, dairy products, water, and other categories are subject to mandatory labeling. Our AI system automates:

• generating a code request in GIS MT;
• receiving codes and applying them to packaging (with position control);
• confirming entry into circulation upon shipment;
• automatic reconciliation—matching codes in the system with physical shipments.

If the discrepancy exceeds 2%, the system blocks shipment and notifies the responsible person. Through automation, we reduce labeling time by 80%.

Comparison of Approaches

Typical Risks and Their Mitigation

Process of Work

1. Analysis and audit — study current processes, data, labeling (2–3 weeks).
2. Architecture design — choose stack: Python, NetworkX, PostgreSQL + DataMatrix generator, ML server (PyTorch/TensorFlow), vector DB for logs.
3. Implementation — develop traceability graph, ML quality model, API for integration with accounting system and GIS MT.
4. Testing — stress test on historical data: recall simulation, response time check (p99 no more than 3 seconds for a 10M-node graph).
5. Deployment and training — deploy on client infrastructure (on-prem or cloud), train operators.

What's Included in the Work

We deliver:

• architecture and API documentation;
• staff training (2–3 days);
• graph and ML model code;
• Power BI dashboards for monitoring;
• SLA for support from 3 months.

Timeline and Cost

Estimated implementation time — from 4 to 6 months, depending on data volume and number of integrations. Cost is calculated individually after audit: considering number of SKUs, shipment points, and required ML accuracy. Get a consultation — we will evaluate your project within 3 business days. Order an audit now to reduce the risk of mass recalls.

We guarantee reduction of problem lot localization time to 15 minutes and reduction of recall volume by 95% compared to a mass approach.

Industry AI Solutions: Healthcare, Finance, Retail, Manufacturing

We encounter the same pain points: a general text model doesn’t distinguish medical nomenclature, and a standard object detector confuses “weld seam scratch” with “casing scratch.” Each time these are different defects with different consequences. To avoid this, we build industry-specific solutions on top of general methods, but with deep domain knowledge — from regulatory requirements to data specifics. Over 5 years, we have completed 80+ projects in fintech, healthcare, retail, and manufacturing, and none were without adaptation to a specific business case.

Healthcare: Regulatory Maze and Data Governance

Medical AI differs not in technical algorithms but in a compliance-first approach. Depending on the country of application, the model may be a Class II or III medical device requiring clinical trials (FDA, CE MDR, GOST R). We ensure compliance with these standards at the architecture stage — fixing them post-factum is 10× more expensive.

Medical imaging. Detection on X‑rays, CT, MRI is a mature area. Models on ResNet, EfficientNet, SegFormer achieve AUC 0.94–0.97 on standard tasks (pneumonia on CXR, polyps on colonoscopy). Key issue is generalization: a model trained on data from one scanner manufacturer degrades on another due to differences in preprocessing and artifacts. Solution: domain adaptation via MONAI (Medical Open Network for AI) from NVIDIA, which includes DICOM loading, 3D augmentation, and confidence calibration. TotalSegmentator — for automatic segmentation of 117 structures on CT, production‑ready, Apache 2.0 license.

Clinical NLP. Extracting structured information from clinical records: diagnoses (ICD‑10/11), prescriptions, dates, indicators. medspaCy, scispaCy, MedCAT — specialized NLP libraries with ontologies (SNOMED‑CT, UMLS). Fine‑tuning BioBERT or ClinicalBERT on our data yields F1 0.85–0.92 on NER tasks versus F1 0.65–0.72 for general BERT. We verified this on a project with a regional oncology center — cancer stage extraction accuracy increased by 23%.

Clinical decision support. LLM assistants for clinical decision support are a regulatory gray area. We use an RAG system on top of clinical guidelines (UpToDate, local protocols) with explicit citation for each statement. The model does not diagnose but helps find relevant protocols. Stack: LlamaIndex + pgvector + pubmedbert-base-embeddings + Llama Guard for safety. Data in DICOM/HL7 FHIR, on‑premise deployment mandatory.

Finance: How to Ensure Interpretability of a Scoring Model under Basel IV?

The financial sector is one of the most mature in applying ML, but regulation is maximal. Every model affecting credit decisions falls under Basel IV, EU AI Act, GDPR Article 22. We deliver AI solutions for fintech that satisfy these requirements — in a project for a top‑10 bank we deployed a scoring model where each record required SHAP explanations.

Credit scoring. Gradient boosting (LightGBM, XGBoost) dominates. Neural networks yield +0.5–2% AUC but lose interpretability. Standard: LightGBM + SHAP to explain each decision. Fairness checking is mandatory: Fairlearn or aif360 for auditing disparate impact on protected attributes (age, gender). The default class is 1–5% — with an imbalance of 1:30, a model with 97% accuracy may have recall 0.2. Solution: focal loss, class_weight='balanced', SMOTE + careful validation. In one fintech scoring project, the model reduced credit losses by $2.1 million annually.

Algorithmic trading and risk management. LSTM and Transformer for price forecasting are popular but unstable in production due to non‑stationarity of financial series. A more robust approach: ML for signal generation (classification: up/down over horizon N) with traditional portfolio optimization on top. Backtesting via Zipline‑Reloaded, vectorbt, QuantLib. Proper backtesting is critical — look‑ahead bias kills results. We guarantee a clean experiment: all data at signal time is available in real time.

AML (Anti‑Money Laundering). Graph Neural Networks for analyzing transaction networks is an actively developing area. PyG, DGL for GNN. Task: detect suspicious patterns in transaction graphs (layering, structuring). Recall is more critical than precision — better 10 false alarms than miss one money laundering. In a project for a large payment service, we increased recall by 18% without increasing false positive rate.

Retail and e‑commerce: Recommendation Systems and Demand Forecasting

Recommendation systems. Current architectural standard: two‑tower model for retrieval + ranking with cross‑features. TensorFlow Recommenders or Merlin from NVIDIA for GPU‑accelerated feature processing. For small catalogs (<100k items), LightFM is sufficient. A common mistake is training on implicit feedback without accounting for position bias. Solution: IPW (Inverse Propensity Weighting) or randomized logging on a portion of traffic. Development time for a basic recommendation system is 4–8 weeks, including A/B test.

Demand forecasting and inventory optimization. Hierarchical forecasting: SKU → category → store → region. HierarchicalForecast from Nixtla automatically reconciles forecasts across levels. TFT or N‑HiTS for base forecast, gradient boosting for adjustment on exogenous factors (promotions, weather, events). One retail project led to a 15% reduction in stock‑outs due to precise promotion calibration.

Visual search and size compatibility. CLIP embeddings for image search — deploy in 2–3 weeks: clip‑ViT‑B‑32 or clip‑ViT‑L‑14, Faiss or Qdrant index, REST API. For size recommendation — specific models on return data and reviews with fit indication.

Manufacturing: Quality Inspection and Predictive Maintenance

Quality control and defect detection. CV models for product inspection are one of the most mature industry tasks. YOLOv10 for defect detection, SegFormer for segmentation. Specifics: class imbalance (defects are rare), high recall requirement (missing a defect is worse than false alarm). Typical dataset: 500–2000 defect images + 500–1000 normal. Few‑shot learning via DINO or SAM 2 works with 50–100 annotated examples. We gained experience on an electronics production line — recall 0.95 at FPR 0.03. A predictive maintenance deployment saved a manufacturing client $500,000 per year in unplanned downtime.

Predictive maintenance. Vibration sensors, current sensors, thermocouples → feature extraction → anomaly or mode classification. Models: LSTM‑AE for unsupervised, LightGBM for supervised (if failure history is available). Integration with SCADA/OPC‑UA via opcua-asyncio or MQTT. Key metric: False Negative Rate — a missed pre‑failure is more costly than a false alarm. Threshold tuned to business cost of each error type. Timeline: 3 to 6 months to production.

Digital twin and simulation. Surrogate models — ML models replacing expensive physical simulation. If a CFD simulation takes 6 hours and a surrogate (trained on 10,000 simulations) takes 0.01 seconds, that's 2,000,000× speedup for optimization. SALib for sensitivity analysis, botorch for Bayesian optimization on top of surrogate.

General Principles of Industry AI

Regardless of industry, there are patterns that work everywhere. Data matters more than architecture. In healthcare, 1000 quality labeled images are better than 100,000 poor ones. In manufacturing, 200 real defect examples are more valuable than 10,000 synthetic ones. Compliance‑first design — regulatory requirements are easier to embed into architecture from the start than to add later. Logging, explainability, versioning from day one. Domain expert on the team — an ML engineer without domain knowledge does slowly and error‑prone what an ML engineer plus a doctor/financier/technologist does quickly and correctly.

We guarantee certification to customer requirements (ISO 13485, SOC 2, GDPR) and provide full model documentation (model card, datasheet, compliance report). Our experience: 10,000+ engineering hours and 80+ projects.

Work Process for an Industry AI Solution

1. Domain immersion (2–3 days) — interviews with experts, studying regulatory requirements, auditing available data.
2. MVP design (1–2 weeks) — stack and architecture selection, feasibility assessment.
3. Development and validation (from 4 weeks to 6 months depending on industry) — model training, testing, compliance.
4. Integration and deployment (1–4 weeks) — on‑premise / cloud / edge, documentation, staff training.
5. Support and monitoring — model drift, retraining, SLA.

Estimated timelines:

Cost is calculated individually for each project. Get a consultation — we will evaluate your dataset, regulatory map, and business goals.

Why Choose Our Industry AI Solutions?

• 80+ completed projects in fintech, healthcare, retail, and manufacturing.
• 5 years on the market — proven experience with compliance and deployment.
• Quality guarantee: we ensure target metrics (AUC, recall, latency p99) and provide full documentation.
• Licensed technologies: PyTorch, MONAI, LightGBM, Qdrant — we use open‑source with commercially safe licenses.
• Flexibility: we work as a contractor or as an extension of your team.

Contact us for a free data audit and consultation. Request a proposal with a detailed work plan. We will discuss your task and prepare a commercial proposal.