-
Notifications
You must be signed in to change notification settings - Fork 0
/
streamlit_app.py
116 lines (81 loc) · 5.21 KB
/
streamlit_app.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
import streamlit as st
import os
#page layout center
# set the page title and icon
st.set_page_config(page_title="Canary Tech", page_icon=" :factory:", layout="centered")
#construct the path of the img folder
img_path = os.path.abspath(os.path.join(os.path.dirname(__file__), 'ressources/img'))
#hide the menu
st.markdown('''
<style>
.stApp [data-testid="stToolbar"]{
display:none;
}
</style>
''', unsafe_allow_html=True)
#center the title
st.markdown("""<h1 style="font-family: Garamond;
text-align: center;">Canary Tech</h1>""", unsafe_allow_html=True)
#make 2 columns
col1, col2 = st.columns(2)
#in col1 the image
canary_path = os.path.join(img_path, 'canary_solo_200.png')
col1.image(canary_path, output_format = "PNG")
#create a title and sub-title for the app
#in col2 put some text
col2.markdown("""<br>""", unsafe_allow_html=True)
col2.markdown("""
##### After years of working in the industrial data engeering field, Otahy has developped a new offer : **Canary Tech** """)
col2.markdown("""##### This concrete portofolio will solve 3 existing pain points of your business in a digital world. """)
st.markdown("""We encounter 3 major recurring pain points with industrial customers:
* 1: Despite all my efforts I have critical orphans data in a "No Tool Land" in between my ERP and other platforms. This creates major risks (quality, finance, compliance, ESG, etc...)
* 2: Everytime I run an analytic project to solve production issues, I start by cleaning, indexing, organising my time series data from machines with a huge associated cost.
* 3: More and more generative AI is beeing used by my employees... officially or not. How I can train my organisation?""")
#orphan data
st.markdown("""## Orphan Data :""")
st.markdown("""They exist because they are small, critical, coming from various sources at different frequencies:""")
st.markdown("""
* Lawyers checking suppliers contracts and compliance then passing instructions to someone else to upload in SAP...
could be scary if you block or enable the wrong company.
* Chief engineer reviewed the product structure and validate the new BOM, but the new BOM is not uploaded in the PDM system.
* Finance receiving recurring late request to validate project margin on Excel with a questionable hourly rate....
(no worries it is just a multi millions call for tender!!) """)
st.markdown("""We have developped a methodology to collect, index, organise, manipulate and secure your orphan process and data in a single place. """)
st.markdown(""":sparkles: We inject the right amount of subject matter expert and IT skills. Resuts are :""")
st.markdown("""* Your processes are captured with the right business rules and experts save times
* IT is finally "taping into" it as they face now a data model with clear rules""")
st.image(os.path.join(img_path, 'orphan_data.png'),output_format = "PNG")
#Times series indexer
st.markdown("""## Timeseries indexer :""")
st.markdown("""Back in a day, you walked into a library and find 1000 books on the same topic. Did you open each one and read it? Or
did you hope to find a book indexer with keywords, synthesis and sample pointing you to the right books?
Same for the Web, you thank google for indexing websites continously.
Why do you not have the same for your time series data from your machines? """)
st.markdown("""Existing technologies allow to go through PetaByte of data and start indexing etc... But at which cost? Here golden rules we learned working with cloud and timeseries :
* It's not because it is possible that it is the right thing to do
* Cloud providers don't crash, they charge
* Don't use serverless architecture when you can plan the workload
* Having IoT monitoring does not mean your data are "ready" for analytics""")
st.markdown("""Thanks to our work with industrial customers on non quality, data integration or predictive maintenance.
:sparkles: We developped a pre processing methodoly to be applied continously on your machine data output, results are:
* Reduce processing cost of accessing your data
* Reduce storage cost applying custom compression
* Increase data quality by continuous cleaning
* Expose critical values, trend, outliners, etc... to your system engineers
* Enable data correlation accross your entire process to support production maturity""")
st.image(os.path.join(img_path, 'sensor_indexer_600.png'),output_format = "PNG")
#Times series indexer
st.markdown("""## Preventive AI Training :""")
st.markdown("""Starting in JAN 2024 ... Stay Tuned
Today, organisation need to learn how to use generative AI and LLM model. Dedicated training will be providing for : """)
col1, col2 = st.columns(2)
col1.markdown("""
* **Everybody**: How interrogate safely a generative AI and validate the output
* **Manager**: How manage the skills recognising the added value of generative AI
* **Executive**: How to make decision with AI, understand conceptually LLM, training dataset and avoid bias""")
col2.image(os.path.join(img_path, 'preventive_ai _400.png'),output_format = "PNG")
st.markdown("""### Contact us for a demo or a POC:
* [[email protected]](mailto:[email protected])
* [[email protected]](mailto:[email protected])""")
#add orphan data image
#st.image(os.path.join(img_path, 'orphan_data.png'),output_format = "PNG")